Você está na página 1de 182


Standard English: The Only Target for 417 Nonnative Speakers of English? Lynn M. Goldstein The Organization of Instruction in Migrant Education: 437 Assistance for Children and Youth at Risk Donna M. Johnson 461 Content and Formal Schemata in ESL Reading Patricia L. Carrell A Cooperative Small-Group Methodology in the 483 Language Classroom Yael Bejarano English Language Proficiency and the 505 Prediction of Academic Success Janet G. Graham Teachers and Students Learning About Compliments Janet Holmes and Dorothy F. Brown
(10-29) To print, select PDF page nos. in parentheses

(30-52) (54-74)

(76-97) (98-114)



Bilingual and ESL Classrooms: Teaching in Multicultural Contexts Carlos J. Ovando and Virginia P. Collier Bilingualism Through Schooling: Cross-Cultural Education for Minority and Majority Students Arnulfo G. Ramrez Reviewed by Constance Walker 556 Computational Linguistics: An Introduction Ralph Grishman Reviewed by Patricia Dunkel CAI Adaptation of Robert J. Dixsons 558 Essential Idioms in English William B. Richardson and Sheldon Wise Reviewed by Vance Stevens 549


Cognitive Style and First Language Background in Second Language Test Performance 565 Lynne Hansen-Strain An Overview of Undergraduate ESL Program Models: A Comparison of Administrative Policies for 570 International Students Elaine Dehghanpisheh 578 The D.C. Schools Project Martha Farmelo

Comments on Elsa Roberts Auerbach's Competency-Based ESL: One Step 583 Forward or Two Steps Back? A Reader Reacts Francis Cartier The Author Responds Elsa Roberts Auerbach 589 Information for Contributors Editorial Policy General Information for Authors 593 Publications Received Publications Available From the TESOL Central Office 595 TESOL Membership Application 608



In This Issue
n The contents of this issue of the TESOL Quarterly reflect dramatically the variety of language teaching issues and learning situations that our profession encompasses. Among the topics covered are the status of standard English as an instructional goal; the organization of instruction in migrant education; the role of content and formal schemata in ESL reading; the effects of small-group learning methods on the acquisition of language skills; the relationship between English language proficiency and academic success in higher education; and cross-cultural research on complimenting behavior as the basis for classroom activities on recognizing and using compliments in English. Lynn Goldstein, who received this years TESOL Research Interest Section/Newbury House Distinguished Research Award at the 1987 TESOL Convention for the study reported in this article, examined the implicit assumptions [of empirical studies of second language acquisition] that the target language group to which the nonnative speaker is exposed is homogeneous and that the target language is synonymous with standard English. The study found that while there was a significant relationship between extent of contact with black Americans and the frequency with which the 28 teenage Hispanic male subjects assimilated grammatical variants of black English, extensive peer contact with blacks was a necessary but not sufficient condition for the acquisition of two features of black English. Goldstein discusses the relevance of her study to research in this area and examines the pedagogical implications of her findings. Donna Johnson reports on a study of the Migrant Education Program, a federally funded program designed to help migrant students in California overcome obstacles to academic success. This large-scale, multisite, multimethod study examined quantitative and qualitative dimensions of the organization of instruction for migrant students and the nature of interactions among program participants. One of the main findings of the study was that an effective partnership between migrant program staff and other staff is a key factor in providing continuity and quality education to migrant students. Like other studies of language minority children, this study suggests that qualified second language specialists . . . need to inform administrators and teachers about the conditions necessary for integrated second language

and intellectual development, and they need to take an active role in becoming involved in the whole-day experiences of migrant students.

Patricia Carrell reports the results of a study which measured the

effects on ESL reading comprehension of both culture-specific content schemata and formal schemata, as well as the interaction of these two types of schemata. Two groups (Muslim and Catholic/Spanish) of high-intermediate ESL students read, recalled, and answered questions about two texts: one with culturally familiar content, one with culturally unfamiliar content. Within each group, half of the subjects read the texts in a well-organized rhetorical format; half read the texts in an altered format. According to Carrell, the overall finding of [the] study seems to be that when both content and rhetorical form are factors in ESL reading comprehension, content is generally more important than form . . . unfamiliar content poses more difficulties for the reader than unfamiliar form. Carrell also points out, however, that each componentcontent and formplays a significant, but different, role in text comprehension. The purpose of the study reported by Yael Bejarano was to assess the effects of two small-group learning methods and of the traditional whole-class method on the general achievement of junior high school EFL learners and on their acquisition of specific language skills. Pretest-posttest comparisons of 665 seventh-grade students in Israel showed that both small-group methodsStudents Teams and Achievement Divisions, a peer-tutoring technique, and Discussion Groupsled to greater gains during the 4-month experimental period than did the whole-class method. According to Bejarano, what promoted general higher language achievement . . . in the classes utilizing small-group techniques was apparently the active communication approachan approach that the author describes in detail. In addition, the two small-group techniques appear to complement one another by serving different teaching objectives. For these reasons, Bejarano recommends the use for language teaching of a cooperative small-group methodology.

Janet Grahams summary of academic prediction studies underscores

the difficulty of using findings to generalize about the relationship of English proficiency to academic success. Graham points out that a number of problemsincluding different criteria for judging academic success, the validity of measures of English proficiency, the means by which data are interpreted, and the large number of uncontrolled variables involved in academic successcontribute to the lack of clear-cut answers for the ESL professional who is looking for guidance in making recommendations to admissions offices. However, while many factors other than English language proficiency are clearly related to academic success and while language test scores should therefore not play a disproportionate role in admissions decisions, it is quite possible that for any institution or program, there


is a minimal level of English proficiency required before other factors assume more importance. Grahams article concludes with suggestions for what ESL practitioners should do when consulted about admissions decisions.

Janet Holmes and Dorothy Brown provide examples of misunder-

standings in compliment exchanges in different cultural contexts and analyze them as instances of pragmalinguistic and sociopragmatic failure. On the basis of an analysis of norms of complimenting behavior in the United States and New Zealand, the authors have developed a number of language classroom exercises, the goal of which is to help learners to develop skill in recognizing compliments, in identifying the appropriate topics and contexts for compliments, and in interpreting their functions appropriately. In arguing their view that sociolinguistic competence should be explicitly dealt with in the language classroom, the authors recognize the need to achieve a balance: While ESL teachers must provide information to learners so that they may choose how to express themselves and do so without unintentionally giving offense, they must not be too prescriptive in terms of what constitutes appropriate linguistic behavior.

Also in this issue: Reviews: Constance walker reviews Carlos J. Ovando and Virginia P. Colliers Bilingual and ESL Classrooms: Teaching in Multicultural Contexts and Arnulfo G. Ramrezs Bilingualism Through Schooling: Cross-Cultural Education for Minority and Majority Students; Patricia Dunkel reviews Ralph Grishmans Computational Linguistics: An Introduction; and Vance Stevens reviews William B. Richardson and Sheldon Wises CAI Adaptation of Robert J. Dixsons Essential Idioms in English.

Brief Reports and Summaries:

Lynne Hansen-Strain reports the results of a study that examined cultural differences in cognitive styles and their effects on second language test performance; Elaine Dehghanpisheh reports the findings of a survey of ESL programs in higher education and describes four undergraduate ESL program models; and Martha Farmelo describes the D.C. Schools Project, which, under the auspices of the Center for Immigration Policy and Refugee Assistance at Georgetown University, operates several tutoring programs for school-age and adult learners of English as a second language. TESOL Quarterly article, Competency-Based ESL: One Step Forward or Two Steps Back? are accompanied by a response by the author. Stephen J. Gaies

The Forum: Francis Cartiers comments on Elsa Auerbachs recent



TESOL QUARTERLY, Vol. 21, No.3, September 1987

Standard English: The Only Target for Nonnative speakers of English?

LYNN M. GOLDSTEIN Monterey Institute of International Studies

An implicit assumption of most research on the acquisition of English as a second language is that standard English is the only target for nonnative speakers of English. The study reported in this article demonstrates that black English served as a target for Hispanic boys acquiring English as a second language in the New York City metropolitan area. In addition, the study shows that extensive peer contact with blacks was a necessary but not sufficient condition for the acquisition of two features of black English, negative concord and distributive be, and that choice of blacks as a reference group played no role in the acquisition of these two features. However, since contact did not account for all the variation and subjects made comments indicating that they had affective responses to input, the study suggests that the concept of reference group should be revised rather than abandoned. White people, they talk different from black, and the teachers, they talk different, you know. (Carlos, a nonnative speaker of English) Carlos and his peers know a lot about language; their knowledge is reflected in comments such as the one above. They know there are many varieties of English in their speech community: Whites, they speak different from blacks (Mario, a nonnative speaker of English). They understand that speakers may vary their English according to setting and interlocutors. In the class, we have to speak nice you know, but not on the street . . . when some people in the street talk bad, you have to speak bad to him (Luis, a nonnative speaker of English). Some even suggest that they have made choices about which variety they want to speak as their second language: I speak like white Americans. Thats a choice (Paul, a nonnative speaker).

Listen to the voices of Carlos, Mario, Luis, and Paul. They may not use the accepted terminology of sociolinguistics or second language acquisition, but they understand the concepts. What is important is that they understand that their speech community is not homogeneous, that they are exposed to many varieties of English, and that learners of English as a second language can make choices among these varieties. Yet, when we review empirical studies of second language acquisition, we observe the implicit assumptions that the target language group to which the nonnative speaker is exposed is homogeneous and that the target language is synonymous with standard English. As a result, except in rare cases (Klein, 1986; Milon, 1975), researchers study the nonnative speakers language in comparison with standard English. In such a framework, forms which deviate from standard English are viewed as either transfer or developmental variants characteristic of the creative processes of second language acquisition. Some deviations from standard English, however, may not be errors at all. Instead, they may be forms characteristic of a different dialect, such as black English. Let us look, for example, at data from two nonnative speakers of English who participated in the study reported in this article: William: When I watch a movie, a scary movie, right, I be you know, I sometimes be thinking of it, you know, how you sometimes be scared. . . . If I tell them a secret, they wont tell nobody else. Paternoster: No, its nothing, because if you got hit by a car, and you dont die, I mean you gonna, you may get hurt for a long time . . . you may be crippled, you know, and thats more dangerous . . . if you are swimming, you may drown . . . you dont feel anything.

William, 16 years old, and Paternoster, 18, were both students at the same high school in the New York City metropolitan area. William had lived in Jersey City for 5, years, Paternoster for 3. Both were born in Ecuador, and both spoke Spanish as their first language. In their speech community they were exposed to many varieties of English, including standard English and black English. William used distributive be, a form characteristic only of black English in which uninflected he is used to indicate qualities which are intermittent rather than permanent; Paternoster never used this form. William used negative concord within the clause, a rule which incorporates a copy of the not in the main verb phrase (VP) or in a preverbal noun phrase onto all indefinite after the main VP. He used this rule categorically (100% of the instances in which it could

be used), similar to speakers of black English, who use negative concord within the clause semicategorically (95%-100% of the instances in which it could be used) or categorically. While Paternoster did not use negative concord in this particular excerpt, an analysis of his data revealed that he used it 62.5% of the time, that is, unlike speakers of black English, who use it 95%-100% of the time. William appeared to have acquired negative concord and distributive be, while Paternoster had not. This is in spite of the fact that they attended the same school, lived in the same speech community, came from the same country, spoke the same first language, and were in the same age group. How, then, do we account for the differences between Williams and Paternosters speech? Why does black English seem to serve as a target for William, but not for Paternoster? Beebe (1985) suggests that second language acquisition (SLA) must view non-native speakers (NNSs) not simply as passive recipients of comprehensible or incomprehensible input from native speakers (NSs) but as active participants in choosing the target language models they prefer. . . . They do not simply acquire a language; they adopt a variety or varieties of that language. (p. 404) Yet almost no research to date has considered the possibility that learners may choose among target language models. Therefore, one goal of the study reported in this article was to demonstrate that nonnative speakers do have target language models other than standard English. We also need to delineate the sociolinguistic factors which might influence whether or not a particular variety will serve as a target. The second goal of the study, therefore, was to explore the relative effects of two factorsextent of contact and feelings of identification with black Americanson nonnative speakers frequency of use of selected features of black English. These factors are most often mentioned as playing a role in the choice by native speakers of a particular variety as a target. Some have suggested that a speakers extent of contact with members of the group who speak the variety may influence the degree to which the speaker uses variants from that groups variety (Labov, 1972a, 1972c, 1973; Reinstein & Hoffman, 1972; Silverman, 1971; Wolfram, 1972, 1973a, 1973b). Others contend that feelings of identification with members of the target language group or that groups covert prestige may influence the degree to which a speaker uses features of that groups language (Hewitt, 1982; Hudson, 1980; Kazazis, cited in Labov, 1972c; Labov, 1972c; LaFerriere, 1979;

LePage, Christie, Turdant, Weeks, & Tabouret-Keller, 1974; Nichols, 1981; Poplack, 1978; Trudgill, 1972). No attempt has been made in second language acquisition research, however, to address the roles these factors might play in the acquisition of a particular variety of English as a second language. Those who have studied extent of contact (Dittmar & Klein, 1978; Mason, 1971; Seliger, 1977; Taylor, Meynard, & Rheault, 1977) or identification (dAnglejan, 1978; Dulay & Burt, 1978; Dulay, Burt, & Krashen, 1982; Ervin-Tripp, 1973; Gardner & Lambert, 1972; Plann, 1975; Schumann, 1978a, 1978b; Spolsky, 1969) do so to further our understanding of why learners achieve different levels of success in second language acquisition; they do not study the roles these factors may play in nonnative speakers acquisition of features from different dialects. THE STUDY Subjects Subjects were selected on the basis of replies to a demographic questionnaire administered by the researcher to approximately 125 students during seven advanced ESL classes at three urban high schools. The demographic questions were read and discussed as the students responded in writing to make sure they all understood the questions. They were asked to supply the following information: age, place of birth, native language, age on arrival in New York City or Jersey City, address, English study prior to arrival (where, when, and how long), length of English study in New York City or Jersey City, and the names of their friends (in the neighborhood and in ESL classes). The answers to these questions were used to eliminate students who did not fit the controls of the study, discussed below, and to assign those who did to friendship pairs. Although 55 students met the criteria for the study, 17 did not participate because they chose not to, could not be paired with a friend, or were unavailable to be interviewed. An additional 10 students were eliminated after being interviewed because they were not proficient enough. In all, 28 students participated in the study. All subjects were nonnative speakers of English. For control purposes they were all males, all native speakers of Spanish, and all from the same age group (15-19, with a mean age of 17.4 years). Males were chosen because they appear to be less sensitive to prestige linguistic norms than are females (Labov, 1972b; Poplack, 1978; Trudgill, 1972). Teenagers were chosen, since researchers claim that they are the most subject to peer pressure and that the 420 TESOL QUARTERLY

vernacular is the most uniform at this age (see Labov, Cohen, Robins, & Lewis, 1968). Teenage subjects also made it possible to compare the results of this study with the findings of previous studies of bilingual teenagers and adolescents (Poplack, 1978; Wolfram, 1973b). Subjects were all proficient speakers of English, as demonstrated by their enrollment in or completion of the most advanced-level ESL course in their respective high schools. The researcher corroborated this assessment during the transcription of the interview data. Advanced students were chosen to avoid, as much as possible, the difficult task of differentiating among linguistic variants which are developmental or transfer variants and those which are variants from dialects other than standard English. All subjects had lived in the New York metropolitan area (New York City and Jersey City) for at least 2 years (mean length of residency = 3.12 years), had arrived as teenagers, and had not studied English before their arrival. Thus, they had acquired English in this area, which was important, since the study focused on the use of variants characteristic of a dialect in this area. The subjects attended three urban high schools, all of which reflected a similar racial and ethnic makeup. Therefore, all subjects had access to peers who were native speakers of the dialects in their speech community. Procedures After the demographic questionnaire had been administered, students were asked to complete measurement instruments for contact and identification during class time. The contact measurement asked subjects to complete a chart on which they named their friends (up to 10) and indicated their ages, ethnicity, race, the language spoken with each, the amount of time spent with each, and the extent of their relationship. The researcher first reviewed the information being requested and examples of a completed chart so that the questions and procedures were understood. She also circulated among the students as they worked, to answer any questions and to check that the chart was being filled out correctly. For the identification measurement, feelings of identification with black Americans were operationally defined as subjects choosing black Americans as their reference group. This decision was based on the work of social psychologists who define reference group as a group with which the individual identifies (Sherif & Sherif, 1964, pp. 54-55) and claim that a reference group becomes

such through an individuals identification with it ( White & Gordon, 1976, p. 21). The instrument used to determine reference group was adapted from Spolsky (1969). A subject rated how well each of 30 descriptors fit his ideal self, black Americans, and white Americans on a 5-point scale ranging from very well to not at all. The subjects filled out the instrument three times: once for black Americans, once for ideal self, and last for white Americans. Before the subjects began, the 5-point scale was translated for them into numerical indexes (very well = almost 100%; somewhat = 75%; average = 50%; only a little = 25%; not at all = 0%). The subjects then were given examples and instructions on how to complete the instrument. Definitions of the descriptors were supplied where needed, and subjects also had written Spanish translations for each descriptor.1 The researcher circulated among the subjects, answering questions, supplying definitions, and checking to make sure that the instrument was being completed correctly. Linguistic data were obtained through interviews of pairs of subjects who were reciprocally named friends. This procedure was intended to minimize the effect of the interview situation (Labov, 1972a) and maximize the effect of peer norms on language (Poplack, 1978; Silverman, 1971). Interviews took place at the subjects schools, but outside of class and in private. Each interview lasted one class period (40-50 minutes) and was audiotaped on a Sony TCS-350 stereo cassette recorder. This recorder allowed the researcher to tape the subjects on separate channels and differentiate between them during the transcription of the interview. The interviews were controlled for content. The researcher chose friendship as the topic for several reasons: (1) It would naturally follow from the questions asked on the demographic questionnaire and contact measurement; (b) it would provide a rationale for being interviewed with a friend; and (c) it would allow the researcher to ask questions about friendships with Americans and about language use, without alerting the subjects to the linguistic purpose of the interview. The primary motivation, however, was that the topic was applicable to the subjects everyday life and would most likely elicit 1 The 30 descriptors (and their Spanish translations) used on the instrument are as follows: busy (atareado), helpful (servicial), sense of humor (iocoso), confident (con confianza en su mismo), competitive (competitivo), broad-minded (con inclinaciones liberales), intellectual (intelectual), optimistic (optimisto), stubborn (cabezudo), kind (benvolo), clever (hbil), efficient (eficiente), considerate (considerado), studious (estudioso), nervous (nervioso), brave (bravo), reasonable (razonable), intelligent (inteligente), friendly (amicable), cool (chvere), sincere (sincero), fashionable (de moda), dependable (confiable), happy (felz), generous (generoso), shy (timido), honest (honesto), good-looking (simptico), popular (popular), calm (calmado).

less formal speech. In addition, woven into the interviews were questions about the danger of death and about fighting, topics which have been shown to elicit vernacular speech (Labov, 1966; Wolfram, 1973b), Role plays concerning interactions between friends were also included in the interview. Again, the primary motivation was that they would minimize the presence of the interviewer and produce vernacular speech (see Poplack, 1978). The role plays were interspersed throughout the interview in a manner that made them appear to be a natural part of the interview. In each case, a discussion led naturally into the role play, enabling the researcher to ask the subjects if they had ever been in a similar situation and providing a purpose for the subjects to enact the role play. The subjects were then given their roles, the situation and roles were described and discussed, and the role play was acted out. Linguistic Variables The linguistic variables of negative concord (within the clause) and distributive be were analyzed as dependent variables. These variables were chosen for the following reasons: (a) The specific variants each group uses and the degree to which each group uses the variants relative to the other group are not changing; and (b) they differentiate between speakers of black English and those of either standard English or nonstandard white English on the basis of how frequently each group uses the variant, ranging from categorically to not at all. They were also chosen because they have been cited in previous studies (Labov, 1973; Labov et al., 1968; Wolfram, 1973b) as differentiating between subjects who have extensive black contact and those who have limited black contact or between those who are integral members of black peer groups and those who are not. Although copula deletion and deletion of final /d/ and /t/ in word-final consonant clusters were also selected as variables, these are not discussed here because subjects did not produce enough tokens of the former and because the analysis of the latter was confounded by processes of second language acquisition (for a full discussion, see Goldstein, 1986). Data Analysis Extent of contact. Extent of contact with black Americans was operationally defined as a function of the number of black friends

a subject named (Silverman, 1971). Thus, for each subject the number of close black friends and the number of black acquaintances were counted, and the amount of time spent with each was determined. This information was obtained from the contact measurement and from information supplied during the interview. Each subject received a score, from 1 to 10, which indicated his extent of black contact. Those with two or more close black friends were placed into the extensive contact group (n = 4) and received scores between 8 and 10. Those with one close black friend and four or more black acquaintances with whom they spent time playing sports or in class were placed into the medium contact group (n = 6) and received contact scores between 5 and 7. Subjects who had no close black friends but who had black acquaintances with whom they played sports or attended classes were placed into the limited contact group (n = 8) and received scores between 2 and 4. Finally, those who had no black peer contacts whatsoever were placed into the no contact group (n = 10) and received a contact score of 1. Reference group. Statistical procedures for finding degree of similarity between groups were used to determine two similarity scores for each subject: one for how similar his ideal self was to white Americans and one for how similar his ideal self was to black Americans. Each subjects reference group was the group which had the lower similarity score. Of the subjects, 5 indicated black Americans as their reference group, and 23 indicated white Americans as their reference group. Linguistic analysis. Each tape was transcribed orthographically in its entirety, and all instances of negative concord and distributive be in the transcripts were underlined. The researcher listened to the tapes a second time and confirmed the transcriptions. Labovs (1966) definition of formality, in which styles can be arranged along a continuum ranging from the most formal style, which occurs when speakers pay the most attention to their speech, to the vernacular, which occurs when speakers pay the least attention to their speech, was employed to analyze speech styles. Casual and careful styles are intermediate, with careful being more formal than casual. For the purposes of this study, casual speech was defined as speech which occurred during the role plays and in response to questions which attenuated the effects of the interview, such as answers to the danger of death question (Labov, 1966; Poplack, 1978). All other speech was defined as careful rather than formal because of the familiarity of the interview topics, the

interview setting outside of the classroom, and the presence of a reciprocally named friend. However, casual and careful speech styles are not differentiated in the analysis, in accordance with Wolfram (1969), because the major concern was to elicit a less than formal style (which both of these are) and because it is not always possible to distinguish between them on the basis of channel clues. After transcription, individual frequency scores were obtained for each subject. Quantitative analysis (see Hudson, 1980, for a complete discussion of this type of analysis) was used to tabulate frequencies for negative concord and distributive be. For the former, the number of times negative concord (within the clause) was used was divided by the number of times it could have been used. For the latter, the number of times uninflected be was used was tabulated. A ratio formula was not used for distributive be for two reasons. First, in addition to distributive be, uninflected be also results when will or would are contracted and then deleted in sentences such as I will be here, yielding I be here. Without a time adverbial, as in I be here next week, it is not possible to determine whether I be here means I will be here or Sometimes I am here. Second, we cannot tabulate be in relation to all present tense conjugated forms of be (is, am, are) because this would involve lumping together two grammatically distinct uses of the verb to be. As a result of the above two factors, Wolframs (1969) suggestion to tabulate distributive be by counting the number of instances of uninflected be was adopted. The score obtained for each of the subjects was used to determine frequency scores for the six groups in the study (extensive contact, medium contact, limited contact, no contact, black reference group, white reference group). To determine group scores for negative concord, the sum of the subjects (in the group) numerators was divided by the sum of the subjects (in the group) denominators. The group frequency scores for distributive be were calculated by adding up the total number of times the subjects in the group used uninflected be. RESULTS AND DISCUSSION Pearson product-moment correlations were carried out to determine if a relationship existed between extent of contact with black Americans and the frequency with which subjects used negative concord and distributive be. The correlations were significant for both (see Table 1).

TABLE 1 Use of Selected Variants

From Black English According to Contact Group Contact group

Linguistic variant Negative concord (NC) Distributive be (DB) *NC = .3682, p <.05.

Extensive (n= 4) 80.00 3.00 *DB = .5544, p <.01.

Medium (n= 6) 67.86 0.33

Limited (n= 8) 49.09 0.00

None (n= 10) 32.69 0.10

One-way analyses of variance (ANOVAS) were employed separately for each of the linguistic variables to determine if the difference among the means of each of the contact groups was significant. The results for negative concord and distributive be were significant (see Tables 2 and 3).
TABLE 2 ANOVA for Negative Concord and Extent of Black Catact

Source of variance
Between groups Witbin groups

11,836.97 26,542.04

df 3 24

MS 3,945.66 1,189.25

F 3.32

* p <.05.
TABLE 3 ANOVA for Distributive Be and Extent of Black Contact

Source of variance Between groups Within groups


s 14.803 7.162

df 3 24

MS 7.321 0.298

F 24.54

p <.01.

Point biserial correlations were employed to determine if a correlation existed between choice of blacks as ones reference group and how frequently subjects used negative concord and distributive be. No correlations were found (see Table 4). The correlations between extent of black contact and frequency of negative concord and distributive be show that contact does play a role in grammatical assimilation. An analysis of individual subjects

TABLE 4 Use of Selected Variants From Black English According to Reference Group

Reference group Linguistic variant Negative concord (NC) Distributive be (DB)

NC =

Black Americans (n=5) 60.00


White Americans (n= 23) 47.50


.13, n.s.

rDB = 2 8 ,


demonstrates, in turn, that extensive or medium contact with black Americans was a necessary, but not sufficient, condition for subjects to use negative concord categorically (that is, like speakers of black English) and for subjects to use distributive be. First, only speakers of black English use negative concord categorically. An analysis of all the subjects reveals that only those subjects who had extensive black contact (William and Enrique) used negative concord in the same way as do speakers of black English (i.e., categorically). All the other subjects used negative concord in a way different from speakers of black English (i.e., variably). On the other hand, 2 subjects who had extensive black contact did not use negative concord categorically; David and Roberto did not behave linguistically like blacks despite having extensive black contact. Thus, only those subjects who had extensive black contact used negative concord categorically, but having this contact did not ensure categorical use. It is, however, possible that those who had extensive contact with blacks but did not use negative concord categorically were not as proficient in English as those who did use it categorically. The interlanguage of nonnative speakers often contains several variants for a target language form. One of these variants is usually the target language form, while the others are not. As the nonnative speaker becomes more proficient in the target language, the frequency of the target language form will increase, and the frequency of the nontarget language forms will decrease. Thus, the nonnative speaker might not use a target language form, such as negative concord, categorically until a high level of proficiency has been attained. Contact was necessary for the assimilation of distributive be, as well. Only 5 subjects, 4 of whom had extensive or medium contact with blacks, used distributive be. However, contact was not a sufficient condition. Of the subjects who had extensive black

contact, one did not use distributive be at all. Pedro, however, was an exception to the claim that extensive contact is a necessary condition for the assimilation of grammatical variants; he had no black contact whatsoever and yet used distributive be one time. It is not at all clear from the data why he did, but given that 17 out of 18 subjects who had limited or no black contact never used distributive be, the evidence is very strong that contact is a necessary condition for grammatical assimilation. These findings about the necessity, but insufficiency, of contact for grammatical assimilation confirm those of Wolfram (1969, 1973b). In his study of teenage Puerto Rican bilingual in New York City, Wolfram (1973b) found that the majority of subjects who used negative concord categorically had extensive black contact and that only those subjects who had extensive black contact used distributive be. In addition, in Wolframs (1969) Detroit study of black native speakers of English, preadolescent and teenage subjects who had predominantly white contact never used distributive be and did not use negative concord categorically. Thus, Wolfram (1973b) states that apparently, it is only through direct peer contact that extensive grammatical assimilation takes place (p. 205). Yet, some of Wolframs New York City and Detroit subjects who had extensive black contact did not use negative concord categorically and did not use distributive be. In discussing the Detroit data, Wolfram (1969) claims that the fact that a UMN [upper middle-class Negro] preadolescent or teenager may have predominantly Negro contacts does not give assurance that one can necessarily expect socially stigmatized grammatical variants (p. 206). While extent of contact with blacks did correlate with the frequency of use of grammatical variants from black English, contact accounts for only 14% of the variation for negative concord (r2 = .1356) and only 31% of the variation for distributive be (r2 = .3074). Contact has also been shown, both in this study and in Wolframs (1973b), not to be a sufficient condition for the assimilation of grammatical variants from black English. This indicates that other factors, in addition to extent of contact, are related to the assimilation of grammatical variants. Clearly, reference group, as it was defined and measured in this study, is not one of these factors, since no correlation was found between reference group and either of the two linguistic variables. This may be because there is no relationship between reference group and grammatical assimilation or because the measurement of reference group was not a valid one. One subject, Roberto,

provides some indication that the measurement may not have really tapped the subjects true reference groups. Roberto scored more similar to white Americans, yet stated during an interview that he felt he had more in common with black Americans. This claim was independently corroborated by his teacher. Comments subjects made during the interviews also indicate that their linguistic reference groups may or may not be the same as their personal reference groups. Many subjects said that they would prefer to speak like white Americans rather than black Americans; a few said that they would prefer to speak like black Americans; and some said that it depended on whom they were speaking with. Paul, for example, when asked whom he spoke English like, answered, The white Americans, and then, when asked whether that was a choice or an accident, said, No, thats a choice . . . because the black Americans talk like whats up. Domingo, on the other hand, said that whom he spoke like depended on whom he was speaking with. He said, I think we should learn to talk different ways . . . when you outside, they talk different ways, so to the right people you talk the right English. These linguistic reference groups may or may not match the subjects general reference groups that were determined by the reference group measure. Paul and John both indicated, for example, that white Americans were their linguistic reference group, as well as their general reference group. Paternoster, on the other hand, implied that whites were his linguistic reference group, but his general reference group was black Americans. In addition, some of the subjects stated choice of linguistic reference group matched their linguistic behavior, and for others it did not. Paul and John, whose linguistic reference group was white Americans, behaved grammatically like white Americans: They never used distributive be, and they used negative concord variably. Enrique, however, said that white Americans were his linguistic reference group but used distributive be and used negative concord categorically; Wilson said that black Americans were his linguistic reference group but never used distributive be and used negative concord variably. In sum, it is not possible to determine why there was no relationship between reference group, as it was measured in this study, and the assimilation of grammatical variants. Conceivably, the instrument employed did not measure reference group, or perhaps reference group is not the same as linguistic reference group. Yet, linguistic reference group, like reference group, did not always match the subjects linguistic behavior.

CONCLUSIONS The discrepancies between what nonnative speakers are exposed to and the language they actually produce has led many to claim that what speakers take in (i.e., intake) is not necessarily equal to what they are exposed to (i.e., input). Most research concerned with this issue has focused on the quantity and quality of the input and interaction and the effects these have on second language development and proficiency (see, for example, Long, 1980). Such research is often based on the premise that standard English is the only source of input and ignores affective responses to input, a third factor that Beebe (1985) and Dulay et al. (1982) claim affects whether or not input becomes intake. This study has demonstrated, through the subjects linguistic behavior and their comments, that nonnative speakers are exposed to more than one source of input and that they do have affective responses to this input. Subjects clearly stated preferences for either black or standard English; and while all the subjects had access to input from black English in their neighborhoods and schools, only some had black English variants in their English. The results of this study have also confirmed Wolframs (1973b) findings that extensive contact with blacks is necessary for the assimilation of grammatical variants from black English. The remarkable similarity between the results of this investigation and those of Wolfram suggests that extensive contact may be necessary for the assimilation of grammatical variants from any groups language. While no relationship was found between reference group and grammatical assimilation, the results for contact indicate that factors other than extent of contact must play a role in grammatical assimilation. The results for reference group may not have been significant because the measurement may not have been valid. Although the choice of blacks as a subjects reference group as defined and measured in this study did not correlate with the use of features of black English, there may be other, as yet unstudied or even undefined, aspects of identification that play a role in grammatical assimilation. Feelings of identification do play a role in language use, language maintenance, and language acquisition; it is logical to assume that they play some role in the acquisition by nonnative speakers of variants from a particular dialect within a speech community. FUTURE RESEARCH Researchers in second language acquisition need to take into account the social context in which their subjects are acquiring

English. The practice in almost all such researchto compare subjects output with that expected of a native speaker of standard Englishignores the fact that speech communities are not homogeneous and that nonnative speakers may have a target other than standard English. Such a practice can result in an inaccurate assessment of second language acquisition. If, for example, a researcher is studying the acquisition of negation, it might be possible to conclude that a subjects categorical use of negative concord over a long period of time means that the subject has not acquired this aspect of English and that the subjects use of negation has fossilized. While this is certainly a possible explanation, it is not the only one. The subjects target may be black English, and in that case, the language has not fossilized. The subject has already acquired negationthe black English pattern. Only by studying the speech community in which second language acquisition takes place, by delineating possible second language targets, and by determining subjects actual targets will researchers obtain an accurate picture of second language acquisition. Future research must also elicit subjects vernacular speech to obtain an accurate measure of their use of variants from black English. Because black English is socially stigmatized, it would hardly be surprising if subjects decreased the frequency with which they used variants from black English, or even eliminated these variants, in an interview situation. Thus, we have no way of knowing why 2 subjects with extensive black contact did not use negative concord categorically and one did not use distributive be. They may not have used them because extensive black contact was not a sufficient condition for them to use these variants or because they reacted to the interview situation or the interviewer. Future research should also study the full range of speech styles nonnative speakers use. Poplack (1978) has demonstrated that people exposed to more than one target may integrate variants from more than one source in a socially appropriate manner. She was able to determine this by eliciting speech from her subjects in more than one speech style. We need to know whether or not nonnative speakers integrate features of more than one dialect into their English and, if so, whether they do this in a socially appropriate way. Whenever possible, ethnographic research methods should be employed. One reason, as mentioned, is that it is important to obtain natural speech samples. Linguistic data obtained by participant-observers in natural settings such as at home, in class, and at play would provide such speech. Another reason is that while

subjects in this study verbally reported their friendships with black and white Americans, a more reliable measure of extent and type of contact might be obtained through actual observations of such contact. The findings of this research are based on data gathered from teenage Hispanic males who lived in the New York metropolitan area. Future research needs to determine if these findings have a wider applicability. We need to study teenage girls, adults of both genders, different ethnic and first language groups, and groups in different geographic areas. Finally, many more variables, in addition to extent of contact and identification, might affect the adoption of a particular linguistic target and the frequency with which the nonnative speaker uses variants from that groups language. These social and psychological variables could include the covert prestige of the target language group, the status of the target language group vis--vis ones own or ones desired status, the difficulty or ease of establishing and maintaining relationships with members of the target language group, the attitudes of ones own ethnic group to the target language group and vice versa, and the instrumental value of using the target language. PEDAGOGICAL IMPLICATIONS Teachers of English as a second language also need to understand that standard English may not be the target for all their students. They need to recognize, for instance, that a student response such as I dont have none may be an error, or it may be the correct use of black English negation. The teachers response to I dont have none, if it is an error, is, of necessity, quite different from the response if it is the correct use of black English negation. While in each case students may need to know the correct standard English form, teachers should give appropriate explanations to the student. In the first case, the student needs to know that what was said was incorrect in standard English; in the second case, the student needs to know that while what was said was correct in black English, there is an alternative in standard English. Furthermore, the student must learn in what contexts the black English version is appropriate and in what contexts the standard English version is appropriate. Teachers, of course, should know what their students second language targets are, if they are to correctly analyze their students linguistic behavior. The findings in this study suggest that this can be accomplished by determining their students friendship patterns with native speakers, by eliciting the names of those they want to

speak like, and by determining in what situations they interact with native speakers. This will not guarantee that teachers will correctly assess their students targets; yet, if a student uses features which could be attributed to black English, has indicated having mostly black American friends, and has shown a desire to speak like them, the teacher would have a strong indication that these features are indeed from black English and are not errors. In the final analysis, however, teachers need to decide what to do if a students target is a nonstandard dialect of English. Certainly, teachers should not take it on themselves to manipulate students friendships or feelings of identification in an attempt to change students targets to standard English. The only recourse teachers have, therefore, is to help their students determine whether or not standard English is necessary to their public lives and to discuss with their students the situations in which they would need to use standard English.

ACKNOWLEDGMENTS The author would like to thank Kathi Bailey, Ruth Cathcart-Strong, and Leo van Lier for their helpful comments on earlier versions of this article, and Leslie Beebe and Frank Horowitz for their comments on the dissertation on which this article is partly based. This article is dedicated to the memory of Bill Mooney, who was instrumental in helping the author locate subjects and who served willingly as a sounding board throughout the study. THE AUTHOR Lynn M. Goldstein is Assistant Professor in the TESOL program at the Monterey Institute of International Studies. Her research interests include sociolinguistics, second language acquisition, and composition. REFERENCES
dAnglejan, A. (1978). Language learning in and out of classrooms. In J.C. Richards (Ed.), Understanding second and foreign language learning (pp. 218-237). Rowley, MA: Newbury House. Beebe, L.M. (1985). Choosing the right stuff. In S.M. Gass & C.G. Madden (Eds.), Input in second language acquisition (pp. 404-414). Rowley, MA: Newbury House.

Dittmar, N., & Klein, W. (1978). The acquisition of German syntax by foreign workers. In D. Sankoff (Ed.), Linguistic variation: Models and methods (pp. 1-21). New York: Academic Press. Dulay, H., & Burt, M. (1978). Some remarks on creativity in language acquisition. In W.C. Ritchie (Ed.), Second language acquisition research: Issues and implications (pp. 65-69). New York: Academic Press. Dulay, H., Burt, M., & Krashen, S. (1982). Language two. New York: Oxford University Press. Ervin-Tripp, S, (1973). Language acquisition and communicative choice. Palo Alto, CA: Stanford University Press. Gardner, R. C., & Lambert, W.E. (1972). Attitudes and motivation in second language learning. Rowley, MA: Newbury House. Goldstein, L. (1986). Linguistic variation in English as a second language as a function of contact versus identification. Unpublished doctoral dissertation, Teachers College, Columbia University, New York. Hewitt, R. (1982). White adolescent Creole users and the politics of friendship. ]ournal of Multilingual and Multicultural Development, 3, 217-232. Hudson, R.A. (1980). Sociolinguistics. Cambridge: Cambridge University Press. Klein, E. (1986). Koreans in Seoul and Honolulu: Aiming for different targets. Unpublished manuscript. Labov, W. (1966). Social stratification of English in New York City. Washington, DC: Center for Applied Linguistics. Labov, W. (1972a). Language in the inner city: Studies in black English vernacular. Philadelphia: University of Pennsylvania Press. Labov, W. (1972b). The reflection of social processes in linguistic structures. In J. Fishman (Ed.), Readings in the sociology of language (PP. 240-251). The Hague: Mouton. Labov, W. (1972c), Sociolinguistic patterns. Philadelphia: University of Pennsylvania Press. Labov, W. (1973). The linguistic consequences of being a lame. Language in Society, 2(1), 81-115. Labov, W., Cohen, P., Robins, C., & Lewis, J. (1968). A study of the nonstandard English of Negro and Puerto Rican speakers in New York City (Final Report, Cooperative Research Project 3288, Vols. 1-2). Philadelphia: United States Regional Survey. LaFerriere, M. (1979). Ethnicity in phonological variation and change. Language, 55, 603-617. LePage, R. B., Christie, P., Turdant, B., Weeks, J. J., & Tabouret-Keller, A. (1974). Further report on the sociolinguistic multilingual communities: Survey of Cayo District, British Honduras. Language in Society, 3 (l), 1-32. Long, M.H. (1980). Input, interaction, and second language acquisition. Unpublished doctoral dissertation, University of California, Los Angeles. 434 TESOL QUARTERLY

Mason, C. (1971). The relevance of intensive training in EFL for university students. Language Learning, 21, 197-204. Milon, J.P. (1975). Dialect in the TESOL program: If you never you better. In M. Burt and H. Dulay (Eds.), On TESOL 75 (pp. 159-170). Washington, DC: TESOL. Nichols, P.C. (1981). Creoles of the U.S.A. In C.A. Ferguson & S.B. Heath (Eds.), Language in the U.S.A. (pp. 69-91). New York: Cambridge University Press. Plann, S. (1975). Acquiring a second language in an immersion classroom. In H.D. Brown, C.A. Yorio, & R.H. Crymes (Eds.), Teaching and learning English as a second language: Trends in research and practice (PP. 213-225). Washington, DC: TESOL. Poplack, S. (1978). On dialect acquisition and communicative competence: The case of Puerto Rican bilinguals. Language in Society, 7(l), 89103. Reinstein, S., & Hoffman, J. (1972). Dialect interaction between black and Puerto Rican children in New York City. Elementary English, 49, 190196. Schumann, J.H. (1978a). The acculturation model for second language acquisition. In R.C. Gingras (Ed.), Second language acquisition and foreign language teaching (pp. 27-50). Arlington, VA: Center for Applied Linguistics. Schumann, J.H. (1978b). The pidginization process: A model for second language acquisition. Rowley, MA: Newbury House. Seliger, H. (1977). Does practice make perfect? A study of interaction patterns and second language competence. Language Learning, 27, 263278. Sherif, M., & Sherif, C.W. (1964). Reference groups. New York: Harper& Row. Silverman, S.H. (1971). The effects of peer group membership on Puerto Rican English. Unpublished doctoral dissertation, Yeshiva University, New York. Spolsky, B. (1969). Aspects of second language learning. Language Learning, 19, 271-285. Taylor, D. M., Meynard, R., & Rheault, E. (1977). Threat to ethnic identity and second language learning. In H. Giles (Ed.), Language, ethnicity, and intergroup relations (pp. 99-118). London: Academic Press. Trudgill, P. (1972). Sex, covert prestige, and linguistic change in the urban British English of Norwich. Language in Society, 1 (2), 179-195. White, R. P., & Gordon, E.E. (1976). Reference groups and significant others: Towards an interface. (ERIC Document Reproduction Service No. ED 140 111) Wolfram, W .A. (1969). A sociolinguistic description of Detroit Negro speech. Washington, DC: Center for Applied Linguistics. STANDARD ENGLISH: THE ONLY TARGET? 435

Wolfram, W.A. (1972). Overlapping influence and linguistic assimilation in second generation Puerto Rican English. In D.A. Smith & R. W. Shuy (Eds.), Sociolinguistics in cross cultural analysis (pp. 15-46). Washington, DC: Georgetown University press. Wolfram, W.A. (1973a). Objective and subjective parameters of language assimilation among second generation Puerto Ricans in East Harlem. In R. Shuy & R. Fasold (Eds.), Language attitudes: Current trends and prospects (pp. 148-173). Washington, DC: Georgetown University Press. Wolfram, W.A. (1973b). Sociolinguistic aspects of assimilation: Puerto Rican English in New York City. Arlington, VA: Center for Applied Linguistics.



TESOL QUARTERLY, Vol. 21, No. 3, September 1987

The Organization of Instruction in Migrant Education: Assistance for Children and Youth at Risk
University of Arizona

The Migrant Education Program is a major provider of instructional services to students whose schooling has been interrupted and who need to learn English as a second language. As part of a statewide evaluation of the program, a study focusing on the organization of instruction and interactions among participants was conducted in 11 school districts in California. A range of qualitative and quantitative data-collection and analysis methods was used. It was found that migrant supplementary instruction, focusing mainly on basic skills, was provided almost entirely by teacher aides. Migrant aides worked in the regular classroom in 68% of the instructional sessions observed and in a pullout setting in 32% of the sessions. Examination of the instructional context of migrant instruction revealed that when students were pulled out of the regular classroom for instruction, the regular teacher showed a very low level of awareness of the migrant students activities or progress in the migrant program. Collaborative planning and strong administrative support for such planning were found to be critical features of high-quality supplemental programs. In the current climate of educational reform, attention is being focused on children and youth who may be at risk of not profiting from school as much as possible or of failing and dropping out altogether (Cook-Gumperz, 1986; Tucker & Mandel, 1986; Tugend, 1986). A major group of students that is considered to be at risk but that has received little attention in the ESL literature is migrant students. This article describes an evaluation of the Migrant Education Program in California which focused on the organization of instruction, the organizational contexts for instruction, and interactions among participants in the process. In general, migrant students in California have at least three characteristics that make them one of the groups of children and

youths at risk. First, migrant students, by definition, have had their schooling interrupted because of moves. For this reason, they often fall behind in school. Migrant students families generally move because the parents are involved in seasonal agricultural and fishing industries. In 1982-1983, for example, about one third to one half of the families of migrant students made at least one move during the school year, while others tended to move during the summer or early fall. Fortunately, many returned to the same town or school district, so that in some cases students were able to resume their schooling with the same teachers. Second, many migrant students are not fluent in the language of the school. Of all third-grade migrant students in California enrolled during 1982-1983, the year in which data were collected, 57% were classified by the state as limited English proficient (LEP). Of those elementary school-age migrant students who enrolled in summer school that year, 46% were classified as LEP. Third, as a group, migrant students academic achievement is low. California migrant students in the third grade (excluding those categorized as non-English-speaking) performed between the 21st and 35th percentile on standardized achievement tests. As is often the case with LEP students in bilingual programs (Douglas & Johnson, 1981), however, their math scores (35) were higher than their English scores (reading, 29; writing, 21). It appeared that those who suffered the most in achievement, as measured by standardized tests, were those students who were most mobile and who were also classified as LEP. Statewide, only about half of all migrant students passed. state proficiency tests in 1982, as compared with about three quarters of the student population as a whole. Moreover, at the upper grades, a lower proportion of migrant students was promoted to the next grade, and 10% fewer graduated from high school than did students as a whole. The Migrant Education Program is a federally funded program designed to assist migrant students in overcoming these obstacles to success in school. In fiscal 1982, the State of California received about $61 million in ESEA (Elementary and Secondary Education Act) Chapter I funds for the program and served 80,000 students from preschool through high school (Noggle, Garza, Weiner, Abrica-Carrasco, & Johnson, 1982). More than three quarters (78%) of these students were in elementary schools. A total of 32,000 migrant students were served in summer school programs. At the time of the study, the program comprised 13 components: overall supplementary instructional services, supplementary secondary instructional services, secondary education pupil personnel services, staff development, needs assessment, utilization

of the Migrant Student Record Transfer System (a computerized system designed to keep track of and transfer student records across states and districts), identification and recruitment, parent involvement, health and supportive services, monitoring and review, interstate and interagency coordination, fund allocation, and information and dissemination. The instructional components were designed to be strictly supplemental to the regular school program and to focus on the basic skills. Because about half of all migrant students in California were acquiring English as a second language, much of the instruction was focused on improving English. Thus, what occurs in migrant education is of concern to the TESOL profession. At the time the study was conducted, there were, however, no specific state policies regarding ESL or bilingual education program models. Policy specified that migrant instruction was to supplement whatever programs the district was providing. The study of classroom practices reported in this article was part of a statewide evaluation conducted over a 2-year period by an educational research corporation under contract to the State Department of Education. It can be described as a multisite, multimethod, large-scale study, employing a range of qualitative and quantitative methods. During the first year, the research staff focused on all 13 components of the program, while in the second year (Noggle et al., 1982, 1983), instruction was one of the central issues investigated, and classroom observations were a major means of gathering data. This report focuses on results of the classroom observations, but because it draws somewhat on findings from the broader, second-year study, the sources and data-collection methods of the overall investigation are briefly described below. For the overall evaluation study, a methodology based on multiple sources of data was used. (See Fetterman, 1986, for discussions of studies employing similar methodologies and Miles& Huberman, 1984, for a discussion of multisite, multimethod, largescale studies. ) Visits were made to each of 10 regional offices and to the State Department of Education, and three 5-day site visits were made to each of 11 districts over the course of the year. On-site interviews were conducted with 19 categories of respondent at the state, regional, district, and school levels, including both nonmigrant and migrant staff. A sample of 100 districts was surveyed by mail to gather questionnaire data on program characteristics as well as data on the English language proficiency, grade advancement, graduation, and mobility of migrant students. Achievement data were also gathered from the statewide testing program, the California Assessment Program. Data obtained from

one source in a particular manner (on-site interviews with aides, for example) were used to cross-validate data obtained from a different source in a different manner (observations of classroom instruction done during the second year, for example) in a continuously interactive system (Noggle et al., 1982). THE STUDY OF CLASSROOM INSTRUCTION During the second year of the evaluation, 86 classroom observations were conducted throughout the state to examine the nature of migrant classroom instruction. In addition to determining who teaches migrant students during migrant supplemental instruction and what they teach, we were primarily interested in describing the nature of the interactions among the various participants in the students education: (a) between migrant staff and migrant students (where instruction is provided, what grouping strategies are used, what general methods are used, and what language use policies and practices are employed); (b) between the regular classroom teachers and migrant students, as related to migrant instruction (amount of interaction, teacher language use, and the extent to which the teacher is aware of the progress the migrant students are making during migrant instruction); and (c) between the migrant aide and the regular classroom teacher (communication, cooperation, evidence of joint planning). Overall we were interested in identifying practices that contributed to highquality instructional environments. Sample From each of the 11 districts to which site visits were made during the second year of the study, three or four schools, those that served the largest numbers of migrant students, were selected, and within each school two classes, on the average, were selected for observation. While a broad range of subject areas and grade levels was observed, 50 of the 86 observations (58%) were of English instruction. Instrumentation The classroom observation instrument, constructed by the research staff to obtain a wide range of information, contained a highly structured section, a semistructured section, and an openended section. The structured section allowed for the collection of data on classroom identity, classroom composition, use of staff with

students, grouping procedures, type of instruction, language use of migrant aide, language use of regular teacher, and amount of communication with migrant students. The semistructured section contained questions dealing with the interactions between staff and students (rapport, teaching methods, engaged time, contribution of instruction to progress in regular class); between regular teacher and migrant students (knowledge of student progress, communication); and between regular teacher and migrant staff (evidence of joint planning, team teaching, rapport). The unstructured section was open-ended to allow for a qualitative, rich description of classroom atmosphere and events. Procedures The observations were conducted by a staff of professional research associates or senior research associates with expertise in the education of language minority students and related areas. All but one had evaluated migrant programs the previous year, and as a result, they were highly knowledgeable of the programs goals, policies, and practices. Training sessions were held prior to site visits in order to clarify category definitions and finalize observation and interview procedures. In addition, while site visits were in progress, the Director of the Study of Instruction contacted site visitors by telephone to resolve any problems, such as lack of clarity regarding the categorization or description of unanticipated circumstances or events. In addition to the observations, both migrant staff and regular (nonmigrant) staff were interviewed regarding instructional practices. Of particular interest was the degree of coordination between migrant staff and the regular classroom teacher. Thus, when migrant instruction was provided in a pullout setting outside of the students regular classroom, the observer went back to the regular classroom to attempt to determine the relationship between the migrant supplementary instruction and regular classroom instruction. In interviews with regular teachers, the extent of coordination was assessed. Coordination was also assessed at the school level through interviews with principals, bilingual program personnel, Chapter I personnel, and other related program personnel. Analysis of Classroom Observation Data Classroom observation data were analyzed using both qualitative and quantitative means. The data from each of the 86 classroom

observations were coded and categorized by two people, the observer and the Director of the Study of Instruction, who also served as the observer at some of the sites. The participation of the Director in every coding session served to provide consistency across observers in the coding of the quantitative data and in the categorizing of the qualitative information. Quantitative data were analyzed using the Statistical Analysis System, a statistical package capable of handling multiple data bases. Analyses consisted primarily of the calculation of frequencies, means, standard deviations, cross tabulations of frequencies, and other descriptive statistics. Qualitative data were analyzed by summarizing information by category and examining the information, along with related information from other sources, for themes or patterns and for examples of exemplary practices. RESULTS Aides as Teachers and Advocates In 99% of the observations, the migrant staff person providing supplemental instruction to migrant students was an aide. While the position of migrant resource teacher was a common one, usually that persons responsibility involved training and supervising aides, rather than teaching migrant students. However, both interviewees and questionnaire respondents indicated that they would have preferred more teaching by certified migrant staff, and indeed there was a trend toward hiring more certified migrant teachers to work directly with migrant students. The programs judged to be of the highest quality by the research staff were those in which certified migrant teachers worked with students. Most of the aides (96%) were bilingual and/or were at one time members of migrant families. Researchers observed and rated the apparent rapport between aides and their students and found that in 98% of the classrooms observed, rapport was fair to excellent. Aides seemed to have empathy for migrant students and were committed both to helping them with their school work and to serving as advocates for them in their total school experience. The extent to which their advocacy was effective depended to a large degree on the contextual factors discussed below. The Content of Instruction Instruction in the Migrant Education Program focused primarily on English, math, and the information needed to pass proficiency tests. Of those programs surveyed by mail, 90% reported providing

ESL instruction, and 40% reported providing assistance with reading and writing in Spanish. Only minor emphasis was placed on supplemental instruction in science and social studies. When such assistance did occur, it was more likely to occur in a bilingual education setting or in a high school. Respondents expressed a need for increased assistance in social studies and science; however, it is likely that in the elementary schools there was not much science instruction to supplement, as this area has been neglected in bilingual education (Horst et al., 1980) and in education in general (Rothman, 1986). Interaction Between Migrant Staff and Migrant Students Setting. Most of the migrant instruction was provided for one class period per day and integrated with mainstream instruction. In two thirds (68%) of the migrant sessions observed in Grades K-12, assistance was provided in the regular classroom, while in 32% of the sessions students were taught in a pullout setting outside of the regular classroom. The pullout groups met in a variety of locations, including resource rooms or laboratories (40%), unused classrooms (30%), designated migrant classrooms (20%), and sometimes even cafeterias (10%). Because ESL teachers are sometimes required to teach in inadequate classroom settings, the research team subjectively assessed and rated the quality of the pullout setting. The pullout location was judged to be nicer than the regular classroom 14% of the time, but 62% of the time it was judged to be not as nice as the regular classroom. Large closet-like settings and noisy cafeterias were among the least desirable settings used. Thus, there were some serious problems with the physical facilities provided to the program for pullout sessions, although these problems are not unusual in supplementary programs. Grouping. Most of the migrant supplementary instruction was provided to students in small groups of two to five (48% of the sessions observed) and to individuals (40% of the sessions observed). Large-group instruction was encountered in only 12% of the observations (see Table 1). Half of the migrant staff who worked with small groups instructed them by pulling them aside in the regular classroom, and the other half met with them outside the regular classroom. Assistance to individuals at their own desks was also a common arrangement that was found in about one quarter of the observations. For example, in high school English and math classes the aide would often sit near the migrant students, listen to

the teachers lecture or instructions to the whole class, and then make sure that migrant students understood and were able to do the assignment.

Location of Instruction by Group Size Group size/location

SmaIt group Futl-aside Pullout

% of time obsetved
24 24 26 11 3 7 5 100

(2 to 5)

Individual At own desk PuMside Pullout Huge group (6 or more) In-class Pullout Totat

Method of instruction. Observers, using fairly standard general educational categories, focused on the types of techniques used by the aides for instruction (see Table 2). The aides spent 46% of their instructional time conducting drill, practice, review, or translation activities. A similar percentage of time (43%) was spent monitoring students work or explaining the work in progress. Only 5% of the time was devoted to the introduction of new material or new concepts, and 4% to discussing or correcting completed work. The aides devoted 1% of their time to each of the following: assigning work or giving directions; administering tests; off-task talk, play, discipline. As Table 2 shows, the nature of the instruction was clearly supplemental. We observed many instances in which the inclass assistance provided, especially at the junior high and high school levels, was highly relevant to helping students succeed in their content classes. Language use. Even though approximately 21% of the states migrant students in Grade 3 were categorized by the state as non-Englishspeaking and an additional 36% as limited English-speaking, there was a strong program-wide, unwritten policy that English rather than the native language was to be used for content instruction to the extent possible, unless the migrant program was supplementing a bilingual education program. This policy was carried out by aides. Although the aides were bilingual and/or former migrants, classroom observations revealed that 26% of them never employed the ESL students primary language in observed instruction. The 444 TESOL QUARTERLY

TABLE 2 Percentage of Time Spent by Migrant Instructional Aides in Various Types of Instructional Activities Instructional activity % of time observed

Note: Percentages do not total to 100 due to rounding.

other 74% used the L1 occasionally (28%), less than half the time (22%), more than half the time (18%), or the entire time (6%). When the L1 was used, it was primarily for reviewing and translating functions rather than for introducing new concepts or material. The unstated language-use policy was as much English as possible and the unstated bilingual education philosophy was transitional. Questionnaire results indicated, however, that 80% of school-district staff were satisfied with the role of the Migrant Education Program in assisting ESL students in the transition from the primary language into English. Although the policy of the program was merely to supplement regular instruction, program staff often took the initiative in establishing an ESL program where there was none; they did not, however, take the initiative in establishing a bilingual program where there was none. Approaches to ESL. Migrant ESL instruction in the elementary schools was most often viewed as a supplement to the standard English language arts program and in some cases as a supplement to the regular teachers ESL program. The most frequently used ESL materials were Quick Start in English, the Defense Language Institute program, and the IDEA Kit. DISTAR, a highly structured, grammatically based oral language program, was in use in some districts. Many Total Physical Response (TPR) workshops were being provided through staff development, and that method was being widely adopted. Several TPR lessons that were observed were carried out in a very mechanical manner, with much emphasis on practicing previously learned commands. ESL was viewed as the application of a structured commercial oral curriculum, rather than as a broad plan for second language

and literacy development that could be implemented throughout the school day in either a bilingual or all-English program. There were few signs of the systematic integration of academic content and language learning in the ESL lessons, and there was no evidence of systematic plans for grouping students and structuring interaction throughout the day in ways that would enhance their second language development. Those ESL activities that were observed indicated little use of authentic materials and few opportunities for students to create their own meanings either in writing or speaking. ln addition to an emphasis on implementing TPR in inservice training, plans were being formulated, with some resistance, to train all regular elementary classroom teachers to provide ESL in the classroom. In sum, most of the ESL instruction I observed was not communicative in nature and was conducted by teacher aides with dedication but limited training. It should be noted that these approaches to providing ESL are not unique to migrant education but are found in many other programs for ESL students. interaction Between Migrant Students and Regular Teachers Contact time with the regular teacher. Educators of language minority students in the fields of ESL, bilingual education, and migrant education have long been concerned about the large amounts of time that students spend being instructed by paraprofessionals rather than by certified teachers. For this reason, researchers tabulated the amount of time regular classroom teachers, migrant aides, or both worked with migrant students during supplemental instruction. The regular teacher and the migrant aide worked together an average of 35% of the class time observed. The aide worked alone with the students an average of 44% of the time, and regular teachers taught alone, usually to the whole class with the aide standing by, an average of 19% of the time. Thus, the upper limit on contact time with the teacher during migrant supplementary instruction was 54% (19% + 35%). When aides were in the regular classroom and the regular teacher and aide were working together collaboratively, instruction was usually closely related to regular classroom activities. Language use by the regular teacher. As expected, regular classroom teachers used the primary language of the migrant ESL students less than did the aides. Only 35% of the regular teachers observed used the primary language of the ESL students. Most of those who

did use the L1 used it not for introducing concepts, but for reviewing or translating material that had been presented in English. This finding is similar to that for the aides and underscores the traditional subordinate role that is played by the non-English language in the classroom. Teacher awareness of student progress. An investigation of the regular teachers knowledge of the progress that migrant students were making during migrant supplementary instruction was undertaken, in the belief that teachers knowledge of student progress is related to student achievement. The issue was investigated by interviewing regular teachers regarding the progress of their migrant students in migrant instruction and rating the extent of their knowledge on a scale of 1 (no awareness) to 4 (considerable awareness). Because these interviews were conducted after researchers had observed the students and were aware of the nature of their supplementary assistance, it was easy to assign a fair and valid rating. Over half the teachers (58%) had considerable awareness of the progress that migrant students were making during migrant instruction. Some 27% evidenced a general awareness, 9% were able to offer no specific information about student progress, and 6% had no awareness at all. Qualitative information from the interviews indicated two reasons for a low level of awareness: (a) The regular teacher and the migrant personnel worked independently of one another, with little communication, and/or (b) the location of migrant instruction in a pullout setting contributed to the lack of awareness. To explore this issue further, the observation data were used to examine the relationship between the regular teachers awareness of student progress and the location of migrant instruction (see Table 3). It was found that only 17% of those teachers who favored smallgroup pullouts were rated as having considerable awareness of their students progress. In contrast, in situations where small-group instruction was provided within the regular classroom, 65% of the regular teachers had considerable awareness of the academic progress of the migrant students. The findings were even more striking for individual instruction; not a single regular teacher who had individual migrant students pulled out was rated as having considerable awareness of their progress. Those employing in-classat-desk and in-class-pull-aside arrangements for individuals were much more knowledgeable about those students; 71% and 86%, respectively, showed considerable awareness of their migrant students progress.

TABLE 3 Regular Teachers Awareness of Progress Cross-Tabulated With Grouping Patterns

These results indicate that pullout supplementary instruction was associated with a disjointedness in the students education. Pullout arrangements as an organizational structure for migrant education, then, may be counterproductive in many situations, particularly if the goal of such instruction is to help students to get along in the mainstream bilingual or all-English curriculum. Often this was a situation over which migrant educators had no control. Because they were to supplement regular instruction, they often could only comply with regular teachers requests about where supplementary instruction should occur. Interaction Between Migrant Staff and Nonmigrant Staff The major finding regarding interaction between migrant and nonmigrant staff, based on data from many sources, was that in most programs there was little evidence of joint planning. Not surprisingly, planning between aides and regular classroom teachers occurred most often when the aide was working in the regular 448 TESOL QUARTERLY

classroom. Clearly, the logistics of joint planning were less difficult to arrange in such situations, particularly if the appropriate administrative support and structure for planning were not otherwise provided. There was also, on the whole, a lack of planning at the program level among the various programs serving migrant students, including Chapter I, bilingual education, and migrant education. Joint planning, however, was found to be a critical ingredient in a high-quality program. Those migrant programs judged as most effective evidenced close communication and coordination among migrant and other staff. The schools that were successful at planning across programs shared two important features of organizational structure: (a) a principal who provided active leadership in promoting planning or a resource teacher who provided the leadership with the support of the principal and (b) the provision of some type of structure for the planning process, such as a committee with regularly scheduled meetings. These features created contexts in which collaboration for the benefit of students was encouraged and supported. DISCUSSION Probably the most significant finding of the overall statewide evaluation study was that while members of the Migrant Education Program staff at all levels were making sincere and dedicated efforts to provide educational continuity for migrant students, an effective partnership between the Migrant Education Program staff and other staff was often missing. This finding was particularly important for instruction. In general, the more migrant instruction was integrated with mainstream bilingual or all-English instruction, the more effective it appeared to be in helping students succeed in school. In viewing the results of this study within the context of current research on second language and literacy development, the conceptual development of language minority students, and supplementary educational programs, a number of recommendations can be made to improve supplementary instructional services to migrant students. The recommendations fall into three general areas: (a) the goals and philosophy of the migrant program, (b) the creation and implementation of what a school district might label its plan for second language and literacy development (as opposed to simply its ESL program), and (c) the organizational structure of supplementary instruction.

Whats Basic? A major goal of the Migrant Education Program is to improve basic skills. The findings of this study suggest that learning basic skills should be interpreted by migrant program staff much more broadly. There are two important considerations here regarding what is basic. First, if the intent of the program is to help students succeed in school, then a focus on basic skills, narrowly defined, can be counterproductive and create additional barriers for many children and youths who may already be at risk. Although prescriptions for effective schools often include more emphasis on basic skills instruction and more time on task (see Bossert, 1985, for a critique), Cazden (1986) and many others have pointed out the problems and consequences of a focus on reductionist concepts of language and learning, especially for younger students. Moreover, children vary in their ability to cope with and profit from poor-quality instruction (Urza, 1986; Wong Fillmore, 1986). A number of researchers and theorists from different fields agree that basic skills and the discrete aspects of language are best learned in the context of meaningful communication about important and interesting topics and tasks (DeAvila, 1985; DeAvila, Cohen, & Intili, 1982; Goodman, 1986; Krashen, 1985; Mehan, Moll, & Riel, 1985; Rigg & Enright, 1986; Wong Fillmore with Valadez, 1986). Second, learning basic skills can be interpreted more broadly to include learning through the languages or varieties the students bring from home, as well as through English, and learning by building on students rich cultural experiences. Migrant instruction in California focused primarily on English, math, and proficiencytest content with a transitional philosophy, and many migrant staff felt that English language development was basic in some sense but that Spanish or Punjabi language development, for example, was not basic. One might argue that allowing children and youths to learn through the languages they bring from home and helping them to develop literacy in the languages used in the home, as well as in the language of the wider environment, are basic in the most fundamental sense of the term. Some research on ESL students indicates that high levels of development of the home language (Cummins, 1986), lots of interaction in the L1 (Neves, 1984), opportunities to discuss concepts in the L1 (Saville-Troike, 1984), experiences in a wide variety of oral and written genres in the L1 (Heath, 1986), and participation in bilingual programs (Ramirez, cited in Crawford, 1986) can contribute not only to achievement in content areas, but to oral and written proficiency in English as well. Thus, migrant

educators should consider contributing to conceptual, language, and literacy development by building more broadly on the linguistic and cultural resources students bring to school (California State Department of Education, 1986). Although no statewide language-use policies existed regarding this issue for the Migrant Education Program, there were at least two reasons for the implicit language-use policy (a) the highly controversial nature of the federal and state role in promoting or even tolerating native language development and instruction (see, for example, Melendez & Lyons, 1986) and (b) the view that native language instruction does not address basic skills. In a context of state laws requiring bilingual education and increasing coordination across programs in staff development programs, migrant staff were, at the time of data collection, becoming informed about and interested in promoting instruction through and development of languages in addition to English, especially for elementary school students. A study conducted in 1987 might find the opposite trend in a sociopolitical context of increased resistance to the use of nonEnglish languages in schools. On the whole, nevertheless, those migrant programs that operated in the context of a bilingual program were rated by the research staff as higher quality programs. In accordance with the Migrant Education Programs emphasis on advocacy, migrant staff could be stronger advocates for the initiation of both high-quality bilingual or multilingual programs and effective ways of promoting second language and literacy development that build onJ students strengths. Migrant students and other bilingual students are at a disadvantage only if their linguistic strengths are underappreciated and schools are failing to build on their strengths (Goodman, 1986, p. 17). A Plan for Second Language Development In addition to eliminating a remedial and reductionist focus on basic skills, other steps can be taken to improve ESL instruction in migrant education. Significant improvements will require that the ESL program be reconceptualized for language minority students in Grades K-12. In migrant education and in bilingual education, the ESL program or component has usually been conceptualized as (a) the application of a structured, commercial ESL curriculum to the students (b) by a designated ESL provider (c) in a situation, defined by time and location, that is identifiable in the organizational structure of the school. This structure for ESL is well entrenched in part because it seems logical to well-intentioned administrators and

in part because it demonstrates compliance with requirements for student services. Current theory and research, however, suggest that the ESL program needs to be conceptualized as a flexible and dynamic plan for second language and literacy development. Unlike the three features listed above, this plan would provide a structure for educators to follow in creating, at various times throughout the day, functional second language learning environments and literate occasions (Graves, 1986) involving the use of the second language. The goals of the plan should call for involving students in authentic written and oral communication (see, for example, Edelsky, 1986; Hayes, Bahruth, & Kessler, 1986) and for creating opportunities for rich language use aimed at the exploration of subject matter and the development of intellectual abilities (Heath & Branscombe, 1985; Milk, 1985; Rigg & Enright, 1986; Saville-Troike, 1984; Widdowson, 1983; Willets, 1986; Wong Fillmore with Valadez, 1986). The development of these goals must be done under the leadership of a qualified professional with expertise in these areas, but in collaboration with others who will be involved in implementation and who are knowledgeable about the broad goals and requirements of multiple programs as well as the overall goals of the school. The design and procedures must be creative and flexible so that the plan can operate in different classroom contexts (bilingual education, ESL, Chapter I, etc. ) and so that resource teachers, teachers, and aides can adapt activities to students changing interests and communicative needs. Chamot and OMalley (1986), for example, have suggested an approach that incorporates a focus on learning strategies (a cognitive academic language learning approach) to promote integrated language use and content learning, while Hudelson (1986) and others have proposed a continual integration of second language and literacy development with content learning. There are numerous ways that opportunities for functional second language use can be created and/or exploited at various times throughout the day, including during integrated thematic units, through content instruction, in the social environment, through classroom management procedures, during the ESL lesson if there is one, and through the use of computers as tools (see, for example, August, 1982; Cathcart-Strong, 1986; DeAvila et al., 1982; Johnson, 1986; Milk, 1985; Miller, 1982). Dialogue journals, learning logs, and peer writing groups are examples of highly flexible techniques that can bridge gaps across instructional settings and address the broad goals of the plan. Migrant educators and

second language educators can serve as better advocates for ESL students by working with others to promote the planning of these kinds of experiences for second language development. The great potential of summer school programs for providing high-quality educational opportunities for migrant students, especially ESL students, should be recognized and exploited. About half of the nearly 32,000 migrant students served in summer school in 1983 were classified as limited in English proficiency. A number of educators expressed the opinion to the research staff that summer school provides excellent opportunities for students to catch up. Because migrant educators are free from many of the constraints and coordination problems of the normal school year, they could explore creative innovations. Thus, summer school offers opportunities to develop innovative second language programs that are part of the overall plan.

Organizational Structure The successful development and implementation of a sound and flexible plan for second language and literacy development make demands of the organizational structure of the school. While much recent work has focused on the relatedness of language and content learning, there has been little discussion of the broader organizational contexts. Only in the wider context of modified organizational arrangements can significant improvements in supplementary instruction for migrant students be achieved. Our study results indicate that an effective partnership between migrant program staff and other staff is a key factor in providing continuity and quality education to migrant students. To make this happen, the school principal must provide strong leadership in supporting the program or must delegate that leadership to a strong resource teacher or qualified specialist who can achieve the types of planning and coordination necessary to carry out an effective and integrated program of supplementary instruction. Joint planning is particularly crucial for pullout programs. The use of pullout instructional arrangements has long been controversial. The temptation is sometimes strong to set up instructional arrangements that demonstrate compliance with external requirements but are not instructionally sound, and the entrenched compliance traditions of the school district often preclude consideration of better instructional arrangements (Knapp & Neufeld, 1985). For example, Kimbrough and Hill (1981) studied the effects of multiple federal programs in school districts that have

trouble administering them. In some of these problem schools, Hispanic migrant students were pulled out for as many as six or seven special sessions per day, and Hispanic students who were learning ESL typically spent half the school day in segregated situations. Similar multiple pullouts were found in bilingual program evaluations (Horst et al., 1980). While problems of this magnitude were not found in this study, the vast majority of regular classroom teachers were clearly not knowledgeable about what their migrant students were doing and what progress they were making in the pullout settings. Many teachers may prefer that ESL and other migrant instruction take place outside of the regular classroom for administrative reasons or because of organizational constraints (Eder, 1986), but the fact that teachers tend to lose track of the content of that instruction and their students progress makes it more difficult to create integrated learning experiences and functional second language learning environments for those students in the regular classroom. A key problem here is the attitude that by providing a few pullout sessions per week ESL is being taken care of. This is not to say that ESL pullout should be eliminated across the board. Indeed, ESL pullout sessions, in some situations, might provide the highest quality experiences of the whole day. Urza (1986) found that this was true for some of the students she observed. Reporting on the educational experiences of several intermediate-grade Southeast Asian children who had been in the United States for 2 years, she commented: The Researcher [Urza] felt as though a match had been struck in a dark room when she realized that the only time the boys ever wrote or spoke of their own experience, the only time they ever used the written word to help themselves know what they could potentially know, the only time they ever had anyone listen and read something which was born of their own special lives was in the ESL classroom. (p. 106) In working with teachers who rely totally on a combination of whole-class and silent-individual-seat-work participant structures (Carter & Canales, 1986; Miller, 1982) or who organize their classes in ways that do not allow for the active participation of second language learners (Enright & McCloskey, 1985; Johnson, 1983), ESL pull-aside and pullout arrangements can be useful. As a plan for second language development is gradually implemented, however, the ESL specialist can work closely with teachers to demonstrate new ways of organizing instruction. Joint planning, then, with administrative support is crucial to design arrangements that are flexible enough to adapt to the varying conditions across classrooms.

The findings of this study, along with those of other researchers who have noted a lack of continuity, a disjointedness in the education of language minority children (Ventriglia, 1982, cited in Cazden, 1986), provide a strong rationale for much more intentional joint planning as well as informal collaboration among the various persons involved in the students education. Discontinuity is greater for migrant students because it exists not only across the school day but also over the months or years as their families move and their schooling is interrupted. Second language educators, in their advocacy for migrant students, should play a strong role in working to reduce this disjointedness. Cazden (1986) has called on ESL teachers, because of their knowledge of how language and literacy are developed and how language can be used to develop intelligence, to play an integrating role by helping to promote deliberately arranged communication among teachers who are teaching the same ESL students. She points out that without such integration, childrens education will be limited not by their abilities or their language, but by the invisible walls between teachers built by categorical funding and separate professional worlds (p. 18). Joint planning can work against migrant students, however, if it contributes to more focus on isolated skills. For examlple, in this study and in an evaluation I recently conducted of a Title VIIfunded ESL program, there were cases in which teachers and ESL aides communicated regularly on an informal basis about students needs. But when teachers (who were not knowledgeable about L2 acquisition) asked aides (who had little training and littke power) to spend more time on spelling exercises and phonics lessons in order to supplement regular class work, the result for the student was more time spent on nonproductive tasks and less time spent on rich learning experiences and the academic discourse forms that lie at the heart of success in the higher levels of schooling (Heath & Branscombe, 1985, p. 3). These kinds of problems can result when seconld language instruction is taken care of by nonprofessionals. Qualified second language specialists, then, need to inform administrators and teachers about the conditions necessary for integrated second language and intellectual development, and they need. to take an active role in becoming involved in the whole-day experiences of migrant students. This is the direction educational reform should take for language minority migrant students, rather than more time on the wrong tasks.



ACKNOWLEDGMENTS This article is a revised version of an invited paper presented at the Colloquium on Classroom Centered Research at the 19th Annual TESOL Convention, New York, 1965. I would like to acknowledge the contributions of the entire research team and to thank the school personnel, students, and parents who participated in the study. I would also like to thank Pat Rigg and three TESOL Quarterly reviewers for helpful comments. THE AUTHOR Donna M. Johnson is Assistant Professor in the Department of English at the University of Arizona. Data were collected for this study while she was Senior Research Associate at RMC Research Corporation, Mountain View, CA. Her research has focused on social factors in second language learning, testing and program evaluation, and the organization of instruction for language minority students.

REFERENCES August, D.L. (1982). The effects of peer tutoring on the second-language acquisition of Hispanic elementary school children. U n p u b l i s h e d doctoral dissertation, Stanford University, Stanford, CA. Bossert, S.T. (1985). Effective elementary schools. In R.M.J. Kyle (Ed.), Reaching for excellence: An effective schools sourcebook (pp. 39-53). Washington, DC: National Institute of Education. California State Department of Education. (1986). Beyond language: Social and cultural factors in schooling language minority students. Los Angeles: California State University, Evaluation, Dissemination and Assessment Center. Carter, K., & Canales, J. (1986). Evaluating second language classrooms: How programs get their effects. Manuscript submitted for publication. Cathcart-Strong, R. (1986). Input generation by second language learners. TESOL Quarterly, 20, 515-530. Cazden, C. (1986). ESL teachers as language advocates for children. In P. Rigg & D.S. Enright (Eds.), Children and ESL: Integrating perspectives (pp. 7-21). Washington, DC: TESOL. Chamot, A. U., & O'Malley, J.M. (1986). A cognitive academic language learning approach: An ESL content-based curriculum. Wheaton, MD: National Clearinghouse for Bilingual Education. Cook-Gumperz, J. (1986). Introduction: The social construction of literacy. In J. Cook-Gumperz (Ed.), The social construction of literacy (pp. 1-15). Cambridge: Cambridge University Press. Crawford, J. (1986, April 23), Immersion method is faring poorly in bilingual study. Education Week, pp. 1, 10. Cummins, J. (1986). Empowering minority students: A framework for intervention. Harvard Educational Review, 56, 18-36.

DeAvila, E. (1985). Motivation, intelligence, and access: A theoretical framework for the education of minority language students. In Issues in English language development (pp. 21-31). Wheaton, MD: National Clearinghouse for Bilingual Education. DeAvila, E., Cohen, E., & Intili, J. (1982). Improving cognition: A multicultural approach. Final report of the MICA Project: Multicultural Improvement of Cognitive Abilities. In S.S. Seidner (Ed.), Issues of language assessment: Foundations and research (pp. 53-81). Chicago: Illinois State Board of Education. Douglas, D. E., & Johnson, D.M. (1981, May). An evaluation of Title VII evaluations: Results from. a national study. Paper presented at the annual meeting of the National Association for Bilingual Education, Boston. Edelsky, C. (1986). Haba una vez: Writing in a bilingual program. Norwood, NJ: Ablex. Eder, D. (1986). Organizational constraints on reading group mobility. In J. Cook-Gumperz (Ed.), The social construction of literacy (pp. 138155). Cambridge: Cambridge University Press. Enright, D. S., & McCloskey, M.L. (1985). Yes, talking! Organizing the classroom to promote second language acquisition. TESOL Quarterly, 19, 431-453. Fetterman, D.M. (1986). Educational evaluation: Ethnography in theory, practice, and politics. Beverly Hills, CA: Sage. Goodman, K. (1986). Whats whole in whole language. Portsmouth, NH: Heinemann. Graves, D. (1986, November). Reading and writing power for teachers for kids: Resources, reflections, beginnings. Keynote address presented at the Teachers Applying Whole Language Conference, Tucson, AZ. Hayes, C. W., Bahruth, R. Y., & Kessler, C. (1986). The dialogue journal and migrant education. Dialogue, 3 (3), 3-5. Heath, S.B. (1986). Sociocultural contexts of language development. In Beyond language: Social and cultural factors in schooling language minority students (pp. 143-186). Los Angeles: California State University, Evaluation, Dissemination and Assessment Center. Heath, S. B., & Branscombe, A. (1985). Intelligent writing in an audience community: Teachers, students and researcher. In S. Freedman (Ed.), The acquisition of written language: Response and revision (pp. 3-32). Norwood, NJ: Ablex. Horst, D. P., Douglas, D. E., Friendly, L. D., Johnson, D. M., Luber, L. M., McKay, M., Nava, H. G., Peistrup, A. M., Roberts, A. O. H., & Valdez, A. (1980). An evaluation of Project Information Packages (PIPS) as used for the diffusion of bilingual projects (Vol. 2) (Report No. VR-460). Mountain View, CA: RMC Research Corporation. (ERIC Document Reproduction Service No. ED 193 954) Hudelson, S. (1986, November). Developing ESL childrens language abilities through the content areas. Paper presented at the meeting of the National Council of Teachers of English, San Antonio. INSTRUCTION IN MIGRANT EDUCATION 457

Johnson, D.M. (1983). Natural language learning by design: A classroom experiment in social interaction and second language acquisition. TESOL Quarterly, 17, 55-68. Johnson, D.M. (1986, November). Using computers to promote the development of English as a second language: A report to the Carnegie Corporation. Tucson: University of Arizona. Kimbrough, J., & Hill, P. (1981). The aggregate effects of federal education programs: Summary. Santa Monica, CA: Rand. Knapp, M., & Neufeld, B. (1985). A proposal for research: A study of the whole-day experience of Chapter I students. Menlo Park, CA: SRI International. Krashen, S. (1985). The input hypothesis: Issues and implications. N e w York: Longman. Mehan, H., Moll, L., & Riel, M.M. (1985). Computers in classrooms: A quasi-experiment in guided change (Final report to the National Institute of Education). La Jolla: University of California, San Diego. Melendez, S. E., & Lyons, J.J. (1986). Testimony of the National Association for Bilingual Education on FY 1987 funding for bilingual education presented to the House Appropriations Subcommittee on LaborHealth and Human ServicesEducation. NABE News, 9 (4, 5), insert 1-10. Miles, M. B., & Huberman, A.M. (1984). Qualitative data analysis. Beverly Hills, CA: Sage. Milk, R. (1985). The changing role of ESL in bilingual education. TESOL Quarterly, 19, 657-670. Miller, W. (1982). Language learning opportunities in bilingual and allEnglish classrooms. In L. Wong Fillmore & S. Ervin-Tripp (Eds.), Sources of individual differences in second language learning (Unpublished report submitted to the National Institute of Education for NIE G-79-0118) (pp. 1-10). Berkeley: University of California. Neves, H.A. (1984). Talking in the classroom and second language acquisition. Unpublished doctoral dissertation, Stanford University, Stanford, CA. Noggle, N. L., Garza, H. A., Weiner, F. S., Abrica-Carrasco, R., & Johnson, D.M. (1982). The education of Californias migrant students: An evaluation of the migrant education programTechnical addendum (RMC Report No. UR-495). Mountain View, CA: RMC Research Corporation. Noggle, N. L., Garza, H. A., Weiner, F. S., Abrica-Carrasco, R., & Johnson, D.M. (1983). The education of Californias migrant students: An evaluation of the migrant education programFinal report ( R M C Report No. UR-495). Mountain View, CA: RMC Research Corporation. Rigg, P., & Enright, D.S. (Eds.) (1986). Children and ESL: Integrating perspectives. Washington, DC: TESOL. Rothman, R. (1986, September 17). Groups foster hands on approach to science in early grades. Education Week, p. 10.

Saville-Troike, M. (1984). What really matters in second language learning for academic achievement? TESOL Quarterly, 18, 199-219. Tucker, M., & Mandel, D. (1986). The Carnegie ReportA call for redesigning the schools. Phi Delta Kappan, 68, 24-27. Tugend, A. (1986, September 17). Youth issues rise in prominence on national agenda. Education Week, pp. 9, 13. Urza, C. (1986). A childrens story. In P. Rigg & D.S. Enright (Eds.), Children and ESL: Integrating perspectives. Washington, DC: TESOL. Widdowson, H.G. (1983). Learning purpose and language use. Oxford: Oxford University Press. Willets, K.F. (Ed.). (1986). Integrating language and content instruction (Educational Report Series). Los Angeles: University of California, Center for Language Education and Research. Wong Fillmore, L. (1986). Research currents: Equity or excellence? Language Arts, 63 (5), 74-81. Wong Fillmore, L., with Valadez, C. (1986). Teaching bilingual learners. In M.C. Wittrock (Ed.), Handbook of research on teaching (3rd ed.) (pp. 648-685). New York: Macmillan.



TESOL QUARTERLY, Vol. 21, No. 3, September 1987

Content and Formal Schemata in ESL Reading

PATRICIA L. CARRELL Southern Illinois University

This article reports the results of an experiment investigating the simultaneous effects on ESL reading comprehension of both culture-specific content schemata and formal schemata, as well as any potential interaction between them. In the study, highintermediate ESL students read, recalled, and answered questions about each of two texts. For each of two groups of readers (students of Muslim and Catholic/Spanish backgrounds), one text had culturally familiar content, the other culturally unfamiliar content. Within each group, one half of the subjects read the texts in a familiar, well-organized rhetorical format, the other half read the texts in an unfamiliar, altered rhetorical format. Results showed the conditions expected to yield good reading comprehension (familiar content, familiar rhetorical form) did so. Similarly, the conditions expected to yield poor reading comprehension (unfamiliar content, unfamiliar rhetorical form) did so. More interestingly, the results for the mixed conditions (familiar content, unfamiliar rhetorical form; unfamiliar content, familiar rhetorical form) indicated that content schemata affected reading comprehension to a greater extent than formal schemata. Specific results are presented and discussed, as are limitations of the study and teaching implications. One type of schema, or background knowledge, a reader brings to a text is a content schema, which is knowledge relative to the content domain of the text. Another type is a formal schema, or knowledge relative to the formal, rhetorical organizational structures of different types of texts. In empirical tests of these two different types of schemata, it is fairly easy to separate out and to test for the effects of one type, while holding the effects of the other type constant. For example, in testing for the effects of content schemata, one keeps the formal rhetorical structure of a text constant, manipulates the content, and has comparable groups of subjects process each different content. Any differences on the dependent measures (answers to literal or

inferential questions, written or oral recall protocols, summaries, and so on) are then presumed to be due to the manipulation of content and readers background knowledge of that content. This type of research has in fact been typical in the field. The seminal study of Steffensen, Joag-dev, and Anderson (1979) is a good example of this type of cross-cultural research on content schemata. In that study, two groups of subjects with different cultural heritages were investigateda group of Asian Indians living in the United States and a group of Americans. Each subject was asked to read and recall two personal letters, both of which were constructed with similar rhetorical organization. However, the cultural content of the two letters differed; one described a traditional Indian wedding, the other a traditional American wedding. It was assumed that all adult members of a society would have a well-developed system of background knowledge about the marriage customs of their own culture and a relative lack of knowledge about the marriage customs of more distant cultures. This is exactly what Steffensen et al. found. Both the Indian and American groups read the material dealing with their own cultural background faster and recalled more of the content. Furthermore, members of the culture provided appropriate cultural elaborations; nonmembers provided inappropriate cultural distortions, frequently outright intrusions from their own culture. In short, the study showed the clear and profound influence of cultural content schemata on reading comprehension. Johnson (1981), who also investigated content schemata while holding formal schemata constant, used two authentic folktales and two groups of readersa group of Iranian students studying in the United States and a group of Americans. Both groups read a Mullah Nasr-el-Din story from Iranian folklore and a story about Buffalo Bill from American folklore. Of the two folktales, Johnson says, both contained similar motifs which were culturally distinct yet were equivalent in plot construction (p. 170). Thus, as with the Steffensen et al. study, Johnson manipulated the content and held the form constant. Johnsons results were much like those of Steffensen et al. superior performance on a given text by members of the cultural group, poorer performance by nonmembers of the cultural groupthus clearly showing strong effects of cultural content schemata. (Johnson, 1982, was omitted from consideration because it was not possible to determine, on the basis of the authors description, whether or not the two experimentally contrived texts were rhetorically equivalent.) Thus, at least these two studies clearly show the separate effects of content schemata, specifically cross-cultural content schemata, on ESL reading.

Likewise, one can test for the effects of formal schemata by keeping the content of a text constant, varying the rhetorical organization, and having comparable groups of subjects process each different rhetorical pattern. Again, one measures differences between the groups on some dependent measure(s) expected to be affected by the differences in comprehension due to the manipulation of form. The same types of dependent measures have been used in this type of schema research: scoring recall protocols or summaries for the number and types of propositions or idea units they contain compared with the original text, or looking at the way different types of literal and inferential questions about the text are answered. At least two different studies of this type have been conducted in ESL reading, one with narrative text (Carrell, 1984b) and one with expository text (Carrell, 1984a). Carrell (1984b) investigated the effects of a simple narrative formal schema on reading in ESL and found differences among ESL readers in the quantity and temporal sequence of their recall between standard and interleaved versions of simple stories. Quantity of recall was enhanced when the storys rhetorical organization conformed to a simple story schemaone wellstructured episode followed by another. When stories violated the story schema, the temporal sequencing of the readers recalls tended to reflect the story schematic order rather than the temporal order of presentation in the story. With expository prose, Carrell (1984a) has shown the effects of four different English rhetorical patterns on the reading recall of ESL readers of various native language backgrounds. Using texts in which identical content information was structured in four different expository patterns, that study showed that the more tightly organized patterns of comparison, causation, and problem/solution generally facilitated the recall of specific ideas from a text more than a more loosely organized pattern called collection of descriptions. There were, however, additional differences among the four native language groups and the four expository text types. Some studies of schematic effects have apparently confounded content and formal schematafailing to distinguish clearly the two types of schemata. (See Carrell, 1983, for a more extensive discussion of the problems of confounding content and formal schemata. ) Kintsch and Greene (1978), who argue that the simple structural story grammars typical of stories of European background may not be typical of stories of other cultural origins, reported differences in the comprehension by American college students of texts of European origin (e.g., Grimms fairy tales) and texts of American Indian origin (e.g., Apache Indian tales). They

concluded from their results that the subjects prior familiarity with the European-based rhetorical organization and their lack of familiarity with the rhetorical organization of the American Indian talesthat is, their formal schematawas the cause of the American students better comprehension of the European texts. But since the texts differed not only in rhetorical organization but also in cultural content, one cannot rule out the possibility that the Americans superior performance on the Grimms fairy tales was due to the more familiar cultural content of those tales or to some combination of formal and content schemata. Finally, one of my own earlier studies (Carrell, 1981), with groups of Japanese and Chinese subjects reading English translations of folktales from their own native culture, as well as from Western European culture and from American Indian culture, suffers from the same potential confounding of content and formal schemata. Although differences in performance related to the cultural origin of the texts were found, it was not possible to determine to what extent these differences were due to content schemata or formal schemata, or to an interaction of the two. Thus, formal and content schemata have been investigated separately and have been confounded in a single study. But to date, no research has been reported which shows the combined effects of both content and formal schemata, in a single controlled study. Although Berkowitz and Taylor (1981) did combine both formal and content schemata in a single study, the design was incomplete. That study investigated the effects of text type (narrative versus expository text) and content familiarity (familiar versus unfamiliar) on native English-speaking sixth graders, but content familiarity was varied only for the expository texts, not the narrative texts. Clearly, prior research on content schemata suggests that texts on content from the subjects cultural heritage, that is, texts with familiar content, should be easier to read and comprehend than texts on content from a distant, unfamiliar cultural heritage. Similarly, research on formal schemata clearly suggests that texts with familiar rhetorical organization should be easier to read and comprehend than texts with unfamiliar rhetorical organization. However, without research on the combined, simultaneous effects of content and formal schemata, no specific predictions can be made about the separate or possibly interactive effects of these two types of schemata. The relative strengths of content and formal schemata in relation to each other are unknown. While previous research leads to the prediction that reading a familiar content in a familiar rhetorical form should be relatively easy and that reading an unfamiliar content in an unfamiliar rhetorical form should be

relatively difficult, no specific hypotheses follow from the previous research about reading a familiar content in an unfamiliar rhetorical form, or about an unfamiliar content in a familiar rhetorical form. The study reported in this article addressed the simultaneous effects of both content and formal schemata in an attempt to formulate and test hypotheses for these conditions. METHOD Subjects This study was conducted with two groups of high-intermediatelevel ESL students enrolled in Levels 4 and 5 in the intensive English program for foreign students at the Center for English as a Second Language (CESL) at Southern Illinois University at Carbondale. TOEFL scores of students in Levels 4 and 5 generally fall in the range of 450 to 525. The subjects included all Level 4 and 5 students who were of Catholic or Muslim religion and who participated in both testing sessions for the study. To obtain large enough sample sizes for each group, the study was run in three separate CESL terms: two terms during the fall of 1985 and one term in the spring of 1986. Group 1 consisted of 28 students of Muslim background, approximately half of whom were Muslim Arabs: 7 from Iran; 4 each from Saudi Arabia and Indonesia; 2 each from Jordan, Palestine, and Malaysia; and 1 each from Egypt, Lebanon, Kuwait, Qatar, Turkey, Pakistan, and Mali. Group 2 consisted of 24 students of Roman Catholic background, predominantly Catholic Spanish speakers from Central and South America: 4 each from Colombia and Panama; 2 each from Brazil, Costa Rica, El Salvador, and Korea; and 1 each from Mexico, Peru, Honduras, Venezuela, Argentina, Malaysia, Poland, and Senegal. Muslim and Catholic were the two cultural groups targeted. For the purposes of this study, religion was considered the defining characteristic of each cultural group. The cultural origin of the texts was also based on religion. This does not mean that for other purposes and in other contexts, religion would necessarily be a stronger determinant of cultural affiliation or identification than, for example, national origin. For example, on nonreligious content, a Catholic from Malaysia may have more in common with a Muslim from Malaysia than with a Catholic from Colombia. (For other studies with religion as the culture-specific determinant, see Lipson, 1983, and Markham & Latham, 1987.)

Materials Texts. The two texts, originally authentic historical biographies of little-known religious personages, were fictionalized in terms of the main character and the events surrounding the main character. This was done to assure that aside from the general cultural-religious knowledge each reader would bring to the text, no one would have specific information about the particular individual or the particular events. Each text carried the name of the fictionalized main character as its title: The Muslim text was entitled Ali Affani; the Catholic text was entitled Saint Catherine. The fictionalized texts preserved the overall rhetorical structure of the original textshistorical narrative. Each text was edited down, however, to two episodes. Therefore, each text began with a historical setting, introducing the main character and setting the time in history. The setting was followed by a two-episode narrative about events in the life of the fictional character. Episode 1 concerned events in the life of the character as a young person; Episode 2 concerned events in the latter part of the characters life. Each text concluded with the death of the character. Thus, these versions were well-formed historical narratives, a common type of rhetorical organization. Furthermore, since the texts each originated from the respective religions/cultures, each represented a rhetorical organization presumed to be familiar to the subjects in the study. To manipulate the rhetorical form of these texts, altered versions were created by interleaving the events from Episode 1 (about the character as a young person) with the events from Episode 2 (about the character as an older person). To assist the reader in keeping track of the different times and/or locations from each episode, the altered texts contained a few additional phrases to help clarify time and place. The texts were considered fair because these phrases were sufficient to enable a careful reader to keep track of the sequence of events surrounding the main character. The settings and conclusions of the texts remained unchanged. Thus, these systematically altered versions of the texts presumably represented an unfamiliar rhetorical organization for this type of historical, biographical narrative. The unaltered and altered rhetorical organizations of these texts (straight temporal order/interleaved order) are presumed to be related to formal schemata the reader has internalized (Mandler 1978a, 1978b; Mandler & Johnson, 1977). The relative ease or difficulty of cognitively processing one rhetorical organization versus another is claimed to be due to, and to be explicable by, the formal schemata of the reader. The readers internalized formal schema for straight temporal sequencing of simple narratives is

what should make processing such texts relatively easy. The absence in the reader of a formal schema for interleaved, or scrambled-order, texts is what should make processing such texts relatively difficult. Hence, it is not the rhetorical structure of the text per se which makes it relatively easy or difficult to process, but how that structure relates to a readers internalized formal schemata. In fact, in the case of both content and form, what we are interested in is the interaction between the content and form of texts and the content schemata and formal schemata, respectively, of readers. (This use of the word interactionto describe an assumed relationship between reader and textshould not be confused with the other way that word is used in this articleto describe a potential statistical relationship between the two independent variables, content and form, in a two-factor research design.) The title of this article is taken from the perspective of the reader, the research design of the study manipulates the text, but the constructs (familiar and unfamiliar content, familiar and unfamiliar form) must be interpreted in terms of the interaction. A text is neither familiar nor unfamiliar, neither easy nor difficult in an absolute sense, but only as a function of the interaction between a reader and that text. Familiar content and familiar form in this study were defined as texts which reflected the content domain of the readers culturalreligious group membership and a well-organized temporal sequence ordering, both presumed to be related to the readers content and formal schemata, respectively. Unfamiliar content and unfamiliar form in this study were defined as texts which reflected a content domain opposite to the readers cultural-religious group membership and an interleaved/scrambled organization, both presumed to be unrelated to the readers content and formal schemata, respectively. In addition, various other formal aspects of the texts were controlled. Both versions of both texts were of approximately equal length (between 250 and 257 words), consisted of approximately the same number of clauses (34 or 35), the same number of T-units (18-22), the same average number of clauses per T-unit (1.6-1.9), and the same average number of words per T-unit (11.4-14.3) (Flahive & Snow, 1980; Hunt, 1965). On the Dale-Chall (1948) readability formula, the texts were determined to be at the fifthsixth grade reading level. Both versions of both texts appear in the Appendix. Multiple-choice questions. A set of 14 multiple-choice comprehension-inference questions was also developed for each text. The

questions were based upon factual information in the text and could not be answered correctly without having read and understood the relevant part of the text. However, each question required the subject to combine the factual information in the text with appropriate inferences. Among the five answer options for each question, the correct choice included culturally appropriate extensions of the information given in the text, and the distracters contained culturally inappropriate extensions of the information given in the text, including some with a probable basis in the opposite culturalreligious group. For example, the Ali Affani text said: Towards the end of the year 405. . . One question on this text said: Alis story most likely took place . . . (a) 405 years after Mohammed left Mecca. [the culturally correct answer] (b) 405 years after the birth of Mohammed. (c) 405 years after Mohammed came to Mecca. (d) 405 B.C. (e) 405 years after the death of Mohammed. It was anticipated that those foreign to the culture of the text but familiar with Catholicism-Christianity might choose the distracter involving B.C. or those involving the birth or death of Mohammed. Such choices might result from schema transfer from the Christian calendar, which dates events relative to the birth of Christ. In another part of the Ali Affani text, Ali is in the sanctuary, AlHaram, in the Holy City, and, the text said, he prayed constantly. One question about this part of the text was as follows: While praying, Ali probably faced . . . (a) Mecca. (b) the Great Mosque. (c) the Kaaba. [the culturally appropriate answer] (d) the west. (e) the east. It was anticipated that those who were unfamiliar with how Muslims pray once they are inside Mecca but who knew something about Muslims facing Mecca or facing the east when they pray would select those distracters. It was anticipated that only insiders to the culture/religion would know the culturally appropriate answer. Debriefing questionnaire. A one-page debriefing questionnaire was developed to elicit relevant information on the subjects country, native language, religion, degree of religiousness (on a 1-10 scale),

degree of prior familiarity with the information in the text (on a 1-5 scale), knowledge of the religion represented by the text (on a 1-5 scale), and ratings of the difficulty of the grammar, vocabulary, content, and overall organization of the text (all on 1-5 scales). Procedures Subjects were tested on 2 successive days in their regular CESL reading classes. Both groups of subjects were exposed to Ali Affani on the first day and to Saint Catherine on the second day. Rather than counterbalance the order of presentation of the two texts, it was decided to control for possible contamination which might result from subjects who were exposed to a text on the first day discussing the content with other subjects who would be receiving that text the next day. Both groups of subjectsMuslims and Catholicsread, recalled, and then answered the questions about each text. Subjects were instructed to read the passage at their own reading rate, not to try to memorize the text but to understand it. They were also informed that they would be asked about the passage later. Recall instructions asked the subjects to write down as much as they could remember from the passage, as exactly as they could, using complete sentences. They could use the words from the texts or their own words, but they were not allowed to refer back to the passage during recall. Within each group, one half of the subjects read the familiar, rhetorically well-organized version, and the other half read the unfamiliar, rhetorically altered version. In addition, between the reading and recall tasks, the debriefing questionnaire served to minimize the effects of short-term memory as well as to elicit relevant information on the subjects. Analyses Data in the study consisted of the answers to the multiple-choice comprehension-inference questions, the variables from the debriefings, and various analyses performed on the written recall protocols. The recall protocols were analyzed for the quantity of idea units recalled from the original text and the quality of the idea unitsthat is, whether the ideas recalled were top-level ideas representing the two central episodes; high-level, or main, ideas within each episode; midlevel ideas, or subtopics; or low-level ideas, or details. Analysis of the reading passages into idea units and into levels of idea units was accomplished by the researcher, with cross-validation by two

research assistants, one familiar with Catholic/Spanish culture/ religion and the other familiar with Muslim culture/religion. In addition, the recall protocols were scored for elaborations and distortions, as well as other errors of recall. Elaborations are culturally appropriate extensions of the text, produced when someone knowledgeable about the culture provides additional culturally correct information not found in or logically inferable from the text; distortions are culturally inappropriate modifications of the text, often outright intrusions from another culture, in which unfamiliar ideas are interpreted, remembered, and recalled in terms of another cultural schema. Because of the cross-cultural nature of the study, every recall protocol was analyzed by two different scorers, one intimately familiar with Catholic/Spanish culture/religion and one intimately familiar with Muslim culture/religion. Reliability between the two judges in scoring the recalls was r = .94. Conflicting scores on the ideas recalled and on elaborations and distortions were resolved by discussion among these two scorers and the experimenter. Statistical analyses were performed with the SAS package of statistical programs on Southern Illinois Universitys IBM 3081-370 computer, using the General Linear Models procedure, which is comparable with analysis of variance. An alpha level of .05 was chosen as the significance level. Nonsignificant results are indicated by n.s.; significant results have the exact probability levels reported. RESULTS Analysis of Debriefing Questionnaires Analysis of subjects responses to the debriefing questionnaire enabled us to check and compare the two groups of subjects. First, there was the expected significant interaction between the groups and their reported degree of knowledge of the religion reflected in the texts, F = 53.62, p = .0001. Subjects were asked to indicate, on a 1-5 scale (1 = nothing, 5 = very much), how much knowledge they had of the religion reflected in the text. The Muslims reported significantly greater knowledge of the Muslim religion (M = 4.11) than of the Catholic religion (M= 1.80), and the Catholics reported significantly greater knowledge of the Catholic religion (M = 3.13) than of the Muslim religion (M= 1.94). Second, when asked to indicate, on a 1-10 scale (1 = not at all religious, 10 = very religious), how religious they felt they were, the two groups were significantly different: Muslims, M = 6.5; Catholics, M = 5.6; F = 4.29, p = .0410. Thus, the Muslims perceived themselves to be more religious than did the Catholics.

There was the expected significant interaction between the groups and their reported degree of prior familiarity with the information in the text, F = 67.36, p = .0001. They were asked to report, on a 1-5 scale (1 = all of it, 5 = none of it), how much of the information in the reading was familiar to them before they read it. The Muslims reported significantly greater prior familiarity with the information in the Muslim text (M = 2.39) than that in the Catholic text (M = 4.04), and the Catholics similarly reported significantly greater prior familiarity with the information in the Catholic text (M = 3.29) than that in the Muslim text (M = 4.04). Subjects also assessed, on a 1-5 scale (1 = very easy, 5 = very difficult), the degree of difficulty of each reading in terms of grammar, vocabulary, content, and overall organization. Subjects assessments of the relative ease or difficulty of content were significantly related to the difference in contentf amiliar versus unfamiliarand subjects assessments of the relative ease or difficulty of overall organization were significantly related to the difference in formfamiliar versus unfamiliar (see Table 1). TABLE 1 Difficulty Ratings as a Function of Familiar and Unfamiliar Form, Content, and the Interaction of Form and Content
Criterion Grammar Form Content Form x Content n.s. n.s. n.s. Vocabulary n.s. n.s. n.s. Content n.s. F = 4.72b p = .0346 n.s. Overall organization F = 4.60b p = .0369 n.s. n.s.

Note: The scale of difficulty was 1 (very easy) -5 (very difficult). a Familiar form M = 2.09, unfamiliar form M = 2.69. b Familiar content M = 1.83, unfamiliar content M = 2.04.

It is noteworthy that this effect, while statistically significant at the .05 level, is not as robust as the other statistical differences found in this study, most of which are significant at .0001. This relatively weak effect on the part of subjects reporting differences in the difficulty of the texts due to content and form may be related to a similar finding by Carrell and Wallace (1983). In that study, neither context nor prior familiarity significantly affected ESL readers ratings of text comprehensibility. Carrell and Wallace concluded

that as a group, the ESL readers tended not to have a sharply honed sense of how easy or difficult a text was for them to understand and that they tended to overrate their comprehension relative to the level of their recall. Texts perceived as equally easy were not recalled equally well. The weak effect here may be due to the same lack of metacognitive sensitivity. Analysis of Question Answers and Recall Protocols Descriptive data for the mean number of questions answered correctly and the mean percentage of idea units recalled from the original texts are reported in Table 2. Even superficial inspection of
TABLE 2 Mean Scores for Question Answers and Quantity of Recall Passage type Question answers Familiar content, familiar form Unfamiliar content, familiar form Familiar content, unfamiliar form Unfamiliar content, unfamiliar form Quantity of recall Familiar content, familiar form Unfamiliar content, familiar form Familiar content, unfamiliar form Unfamiliar content, unfamiliar form M 6.58 3.38 6.08 3.62 43.35 35.73 39.58 33.15

these data shows that content was a stronger predictor of performance than was form or any interaction between the two. And indeed, inferential statistical analysis confirms that impression (see Table 3). In other words, based on these results, not only are
TABLE 3 Question Answers and Quantity of Recall as a Function of Familiar and Unfamiliar Form, Content, and the Interaction of Form and Content

Question answers
Form Content Form x Content

Quantity of recall
n.s. F = 13.65 p = .0005 n.s.

n.s. F = 64.43 p = .0001 n.s.

the predictions of the original hypotheses confirmed (familiar content-familiar form = easy; unfamiliar content-unfamiliar form = difficult), but it would appear that reading familiar content even in an unfamiliar rhetorical form is relatively easy, whereas reading unfamiliar content even in a familiar rhetorical form is relatively difficult. However, lest we conclude too hastily that the rhetorical form of a text has no effect on ESL reading comprehension, when familiarity of content is also a factor, the analysis of the type of information, or the kinds of idea units, recalled revealed one very interesting, significant effect of form (see Table 4). When recalls
TABLE 4 Type of Idea Units Recalled as a Function of Familiar and Unfamiliar Form, Content, and the Interaction of Form and Content Type of idea unit

were scored for whether they clearly expressed the two top ideas of each text, that is, the two central and separate episodes in each text, the form of the text was found to be a significant factor. Subjects who read the versions of the texts in which the events from the two episodes were interleaved failed to express clearly the separateness of these two episodes in their recalls. It appears that these subjects did not clearly understand that each text was about two separate time periods in the life of the main character. Events from the second episode, the latter part of the characters life, were confused with events from the first episode, the early part of the characters life. Subjects in this condition tended to recall the text as one single episode and had no sensitivity to the time differences of the events. Table 4 also shows a significant main effect for familiarity of content at the high level of idea units, that is, at the level of main ideas within each episode. Subjects familiar with the content of a text recalled significantly more main topics and major idea units

from each episode than did subjects unfamiliar with the content. Familiar versus unfamiliar content and form had nonsignificant effects on the recall of midlevel idea units (subtopics within each episode) and of low-level idea units (details). The results of scoring culturally appropriate elaborations and culturally inappropriate distortions and other errors are shown in Table 5. Superficial inspection of the data in Table 5 indicates that
TABLE 5 Mean Number of Idea Units Elaborated or Distorted per Recall Protocol Passage type Elaborations Familiar content, familiar form Unfamiliar content, familiar form Familiar content, unfamifiar form Unfamiliar content, unfamiliar form Distortions Familiar content, familiar form Unfamiliar content, familiar form Familiar content, unfamiliar form Unfamiliar content, unfamiliar form M 2.35 0.19 2.42 0.31 0.08 3.38 0.15 3.15

content is a stronger source of elaborations and distortions than form, and indeed inferential statistical analysis confirms this impression (see Table 6). Those familiar with the cultural-religious background of the text engaged in culturally appropriate elaborations in their recall protocols; those unfamiliar with the cultural-religious background of the text made culturally inappropriate modifications of the text, including obvious intrusions from their own cultural-religious background. Whether the form was rhetorically familiar or unfamiliar had no significant effect on subjects engaging in culturally based elaborations or distortions.
TABLE 6 Elaborations and Distortions as a Function of Form, Content, and the Interaction of Form and Content Elaborations Form Content Form x Content

Distortions n.s. F = 80.43 p= .0001 n.s.


n.s. F = 65.80 p= .0001 n.s.

Examples of culturally appropriate elaborations provided by insiders to the culture/religion include the following from a Muslim on the Ali Affani text he went to Mecca because he wanted to pray in Al Haram, because when you pray one time in Mecca its equal one thousand than other place. The following elaborations were provided by Catholics on the Saint Catherine text: She wanted to join the Dominican Order, a religious order, She carried for the sick in a Catholic hospital, when she was 21 she became a nun, and Catherine died at 33 years old, the same age than Christ. One Catholic even ended his recall protocol with the word Amen. Examples of culturally inappropriate distortions by those outside the culture/religion of the text include the following from Catholics on the Ali Affani text: when he was walking to the church and he went to live in a monastery. The following was typical of the distortions provided by Muslims on the Saint Catherine text: She lived in Italy with her holy family. Finally, recall protocols were scored for three different kinds of errors: (a) sequence errors, that is, the number of idea units which were involved in distortions of the chronological sequence of events; (b) blends, that is, the number of pairs or sets of idea units blended together in recall; and (c) other errors, including lexical errors, syntactic errors, and errors whose source was otherwise inexplicable. The third category included such errors as a subjects recalling that Ali Affani found his mother in a tree or in a chair, rather than in the street, or recalling about St. Catherine that she didnt nurse sick instead of when she was not nursing the sick. As can be seen in Table 7, no significant effects were obtained for blends or other errors. Only sequence errors showed significant effects, and these were due to form. As previously mentioned, those
TABLE 7 Errors as a Function of Form, Content, and the Interaction of Form and Content

Form x Content




Familiar form M = 0.62, unfamiliar form M = 6.00).



who read the unfamiliar, rhetorically altered forms of the texts tended to confuse sequences of events and relationships between events. For example, in the Saint Catherine text, Catherine told her parents that she wanted to become a nun, and did become a nun, before she had a vision. Readers of the unfamiliar, rhetorically altered form often recalled her telling her parents she wanted to become a nun after she had had her vision. DISCUSSION The overall finding of this study seems to be that when both content and rhetorical form are factors in ESL reading comprehension, content is generally more important than form. When both form and content are familiar, the reading is relatively easy; when both form and content are unfamiliar, the reading is relatively difficult. When either form or content is unfamiliar, unfamiliar content poses more clifficulties for the reader than unfamiliar form. However, perhaps not too surprisingly, rhetorical form is a significant factor, more important than content, in the comprehension of the top-level episodic structure of a text and in the comprehension of event sequences and temporal relationships among events. In other words, each componentcontent and formplays a significant, but different, role in the comprehension of text. Further research on the combined effects of content and form in ESL reading comprehension is clearly needed. This study is a first of its kind and involved only one particular manipulation of content and only one particular manipulation of form, with rather specific cultural groups of ESL readers at one proficiency level (highintermediate). Further studies are needed to examine other proficiency levels, other cultural groups, and other types of manipulation of content and form. For example, other kinds of manipulations of a texts rhetorical organization, related to the formal schemata of various types of readers, may yield results different from those obtained in this study. Failing this additional research, the teaching implications of this study should not be overstated. However, I do not believe it would be an overstatement to say that this study suggests, as have others before it (for example, Carrell& Eisterhold, 1983), that in the ESL reading classroom, content is of primary importance. As Steffensen et al. (1979) have said, the schemata embodying background knowledge about the content of a discourse exert a profound influence on how well the discourse will be comprehended, learned, and remembered (p. 19). Teachers of ESL reading need to be aware of the important role

in ESL reading of background knowledge of text content, especially cultural content, and they must often be facilitators of the acquisition of appropriate cultural content knowledge. Stevenss (1982) observation about L1 reading teachers applies equally, if not more so, to ESL reading teachers: A teacher of reading might thus be viewed as a teacher of relevant information as well as a teacher of reading skills (p. 328). In addition, however, as I have suggested elsewhere (Carrell, 1985) based on related research, ESL reading teachers also need to be cognizant of the rhetorical organization of texts and should teach students to recognize and use the top-level rhetorical organization of text to facilitate comprehension and recall.

ACKNOWLEDGMENTS I wish to thank my research assistants, Jo Ellen Rayyan and Aida Prez, for their assistance with various aspects of this project. I also thank my colleagues in the Center for English as a Second Language who facilitated my access to the international students who took part in the study. Further, I would like to thank two colleagues, Sandra Silberstein and Ulla Connor, for reading and commenting on an earlier draft of this article; they bear no responsibility for any remaining difficulties. Finally, this article has benefited from the comments of two anonymous reviewers, to whom I am grateful. This research was partially supported by an internal grant from the Office of Research Development and Administration at Southern Illinois University at Carbondale. THE AUTHOR Patricia L. Carrell is Professor of Linguistics/ESL and Psychology and Associate Dean of the Graduate School at Southern Illinois University. Her previous research on content and formal schemata and ESL reading, for which she was awarded the 1985 Paul Pimsleur Award for Research in Foreign Language Education by the American Council on the Teaching of Foreign Languages, has been published in the TESOL Quarterly, Language Learning, and The Modern Language Journal and will also appear in forthcoming volumes co-edited with Dave Eskey and Joanne Devine. REFERENCES Berkowitz, S., & Taylor, B. (1981). The effects of text type and familiarity on the nature of information recalled by readers. In M.L. Kamil (Ed.), Directions in reading: Research and instruction (pp. 157-161). Washington, DC: National Reading Conference.

Carrell, P.L. (1981). Culture-specific schemata in L2 comprehension. In R. Orem & J. Haskell (Eds.), Selected papers from the Ninth Illinois TESOL/BE Annual Convention, First Midwest TESOL Conference (pp. 123-132). Chicago: Illinois TESOL/BE. Carrell, P.L. (1983). Some issues in studying the role of schemata, or background knowledge, in second language comprehension. Reading in a Foreign Language, 1, 81-92. Carrell, P.L. (1984a). The effects of rhetorical organization on ESL readers. TESOL Quarterly, 18, 441-469. Carrell, P.L. (1984b). Evidence of a formal schema in second language comprehension. Language Learning, 34, 87-112. Carrell, P.L. (1985). Facilitating ESL reading by teaching text structure. TESOL Quarterly, 19, 727-752. Carrell, P. L., & Eisterhold, J.C. (1983). Schema theory and ESL reading pedagogy. TESOL Quarterly, 17, 553-573. Carrell, P. L., & Wallace, B. (1983). Background knowledge: Context and familiarity in reading comprehension. In M.A. Clarke &J. Handscombe (Eds.), On TESOL 82 (pp. 295-308). Washington, DC: TESOL. Dale, E., & Chall, J.S. (1948). A formula for predicting readability: Instructions. Educational Research Bulletin, 27, 37-54. Flahive, D. E., & Snow, B.G. (1980). Measures of syntactic complexity in evaluating ESL compositions. In J. W. Oller, Jr., & K. Perkins (Eds.), Research in language testing (pp. 171-176). Rowley, MA: Newbury House. Hunt, K. W. (1965). Grammatical structures written at 3 grade leuels. Champaign, IL: National Council of Teachers of English. Johnson, P. (1981). Effects on reading comprehension of language complexity and cultural background of a text. TESOL Quarterly, 15, 169-181. Johnson, P. (1982). Effects on reading comprehension of building background knowledge. TESOL Quarterly, 16, 503-516. Kintsch, W., & Greene, E. (1978). The role of culture-specific schemata in the comprehension and recall of stories. Discourse Processes, 1, 1-13. Lipson, M.Y. (1983). The influence of religious affiliation on childrens memory for text information. Reading Research Quarterly, 18, 448-457. Mandler, J.M. (1978a). Categorical and schematic organization in memory (Tech. Rep. No. 16). San Diego: University of California, Center for Human Information Processing. Mandler, J.M. (1978b). A code in the node: The use of a story schema in retrieval. Discourse Processes, 1, 14-35. Mandler, J. M., & Johnson, N.S. (1977). Remembrance of things parsed: Story structure and recall. Cognitive Psychology, 9, 111-151. Markham, P., & Latham, M. (1987). The influence of religion-specific background knowledge on the listening comprehension of adult secondlanguage students. Language Learning, 37, 157-170. 478 TESOL QUARTERLY

Steffensen, M. S., Joag-dev, C., & Anderson, R.C, (1979). A cross-cultural perspective on reading comprehension. Reading Research Quarterly, 15, 10-29. Stevens, K. (1982). Can we improve reading by teaching background information? Journal of Reading, 25, 326-329.

APPENDIX Saint Catherine (Familiar, Well-Organized Form) About six hundred years ago, there was a young woman named Catherine. She lived with her parents in Italy. As a young child, because she was born on the Feast of the Annunciation, Catherine loved the Blessed Mother and the Holy Family very much. Whenever she climbed up or down stairs, she would kneel on each step and say a Hail Mary. She never went anywhere without her rosary. When she was sixteen years old, a rich man was eager to marry Catherine. Her parents liked him and wanted her to marry him, but Catherine did not want to. After refusing to marry the rich man, Catherine told her parents she wanted to become the bride of Christ. Her parents were angry, but they finally agreed. When she was twenty-one years old, Catherine joined the Dominican order. She worked in the hospitals with the other young women of the order. When she was not nursing the sick, she was at Mass. One day, on Ash Wednesday, after receiving communion as a Dominican, Catherine had a remarkable vision. While looking at the cross, five blood-red rays of light came from the cross and touched her hands, feet, and chest. After her vision, Catherine fainted. She got better quickly, but the scars remained on her body for the rest of her life. This was not long, however; she died when she was thirty-three. Because of this, and other remarkable things about her life, she is known today as Saint Catherine. Saint Catherine (Unfamiliar, Altered Form) About six hundred years ago, there was a young woman named Catherine. She lived with her parents in Italy. As a young child, because she was born on the Feast of the Annunciation, Catherine loved the Blessed Mother and the Holy Family very much. Whenever she climbed up or down stairs, she would kneel on each step and say a Hail Mary. She never went anywhere without her rosary. When she was twenty-one years old, Catherine joined the Dominican order. She worked in the hospitals with the other young women of the order. When she was not nursing the sick, she was at Mass.

When she was sixteen years old, a rich man was eager to marry Catherine. Her parents liked him and wanted her to marry him, but Catherine did not want to. One day, on Ash Wednesday, after receiving communion as a Dominican, Catherine had a remarkable vision. While looking at the cross, five blood-red rays of light came from the cross and touched her hands, feet, and chest. After refusing to marry the rich man, Catherine told her parents she wanted to become the bride of Christ. Her parents were angry, but they finally agreed. After her vision, Catherine fainted. She got better quickly, but the scars remained on her body for the rest of her life. This was not long, however; she died when she was thirty-three. Because of this, and other remarkable things about her life, she is known today as Saint Catherine. Ali Affani (Familiar, With &@zed Form) There once was a young man named Ali Affani. He lived in Jidda with his widowed mother. Towards the end of the year 405, young Alis mother agreed that he could go to Mecca as all good men do. While in the desert, on his way to Mecca, something happened which made young Ali unfit to continue his trip. Believing that his trip had begun badly, he returned to Jidda. Upon returning to Jidda, young Ali found his mother sitting in the street, crying and tearing her clothes and hair like a crazy woman. She told Ali that since he had left, she had been in the street. She would not enter the house without her son. Ali really wanted to go to Mecca, but could not leave her sitting outside, so he stayed home. Ali was finally able to go to Mecca, several years later, after his mother died, in the year of 420. Ali spent the rest of his life in the sanctuary, AlHaram, in the Holy City. He only left once each day to buy food. He did not need to buy water because God provided it. As an old man, Ali was very religious and prayed constantly. Each day he would read from the Koran while walking around the Kaaba a number of times. In the sanctuary in Mecca, whenever Ali walked around the Kaaba, he would show his respect for the precious stone. Ali died in the Great Mosque, his home for thirty years. Ali Affani (Unfamiliar, Altered Form) There once was a young man named Ali Affani. He lived in Jidda with his widowed mother. Towards the end of the year 405, young Alis mother agreed that he could go to Mecca as all good men do. Ali was finally able to go to Mecca, several years later, after his mother died, in the year of 420. Ali spent the rest of his life in the sanctuary, Al-Haram, in the Holy City. He only left 480 TESOL QUARTERLY

once each day to buy food. He did not need to buy water because God provided it. While in the desert, on his way to Mecca the first time, something happened which made young Ali unfit to continue his trip. Believing that his trip had begun badly, he returned to Jidda. As an old man in Mecca, Ali was very religious and prayed constantly. Each day he would read from the Koran while walking around the Kaaba a number of times. Upon returning to Jidda, young Ali found his mother sitting in the street, crying and tearing her clothes and hair like a crazy woman. She told Ali that since he had left, she had been in the street. She would not enter the house without her son. Ali really wanted to go to Mecca, but could not leave her sitting outside, so he stayed home. In the sanctuary in Mecca at last, whenever Ali walked around the Kaaba, he would show his respect for the precious stone. Ali died in the Great Mosque, his home for thirty years.



TESOL QUARTERLY, Vol. 21, No. 3, September 1987

A Cooperative Small-Group Methodology in the Language Classroom

Everymans University, The Open University of Israel

The study reported in this article assessed the effects of two small-group cooperative techniques (Discussion Group; Student Teams and Achievement Divisions) and the whole-class method on academic achievement in EFL for 665 pupils in 33 seventhgrade classes. The students were taught by 18 teachers (assigned at random to one of the three methods), who participated in training workshops followed up by in-class coaching. Evaluation of pupils achievement was conducted by observation and by special achievement tests administered before and after the experiment. Particularly noteworthy are the findings revealing that both group methods registered significantly greater improvement than the whole-class method on the total score of the test and on the listening comprehension scale. These findings support the link between the communicative approach to foreign language instruction and cooperative learning in small groups. The study demonstrates how to forge a link between the content and the process of instruction.

The emphasis in language teaching has recently shifted from purely structural competence to communicative competence from the ability to merely manipulate the linguistic structures correctly to the ability to use the language appropriately in real communication. It is generally agreed that there should be a balance between grammatical accuracy and communicative effectiveness (Littlewood, 1981), between use and usage (Widdowson, 1978). In Hymess (1972) words, There are rules of use without which the rules of grammar would be useless (p. 275). Thus, it is reasonable to predict that language instruction which emphasizes interpersonal communication will be more effective than instructional methods concentrating on having students learn rules of grammar (Breen & Candlin, 1980; Brumfit, 1980; Brumfit & Johnson, 1979; Krashen, 1981; Littlewood, 1981). The best classroom lessons in English are therefore those in which pupils

understand what is said to them and in which they are directly and actively involved in a natural process of communication with others. Along this line of thought, Littlewood (1981) has suggested the incorporation of three perspectives in the communicative curriculum: structural, functional, and communicative. The structural perspective involves the rules of grammar, and the functional perspective is the way structure is related to its communicative functions (e.g., note the use of the modal may in asking for permission in the sentence, May I borrow your pen?); these two perspectives can be taught formally. The communicative perspective, which is the appropriate use of language as an instrument for social communication, is acquired informally by actually participating in such communication. This perspective entails a more informal and natural acquisition of strategies for communication, which include the ability to use the language system creatively and to relate forms to function appropriately in situations involving real meaning, real time, and actual interaction. These three perspectives, as outlined by Littlewood, will enable learners to communicate appropriately in the new language. They illustrate the linguistic content the communicative syllabus aims to attainor the what. But the problem for designers and providers of language instruction has been to determine how to apply this new approach in the language classroom (Johnson & Morrow, 1981; Krahnke & Christison, 1983; Krashen, 1981; Stevick, 1980). A methodology is needed which will bridge the gap between the language content we teach (the what) and the instructional methodology (the how). A few attempts have been made by methodologists like Curran (1972, 1976), Gattegno (1972, 1976), Lozanov (1979), Asher (1977a, 1977b), and Winitz and Reeds (1973), but they are hardly applicable on a large scale. Moreover, none of these methodologists has drawn any systematic conclusions on the results of broad-based research in applying their methodologies. This article reports the results of a study designed to assess a cooperative-group methodology that combines the what and the how in the language classroom and was thus intended to answer many of the pedagogical questions ESL/EFL teachers are faced with as a result of the content/process dichotomy discussed above. In effect, two levels of theoretical principles guided this study. One relates to the instructional contentbased on theories of language and language acquisition. The other relates to the psychological dimension of classroom life, to the problems of human interaction in the small-group social setting and to organizing task-oriented groups of pupils so that they can function productively.

One of the most prominent bodies of research and practice concentrating on changing the traditional classroom in order to improve learning and social relations among classmates is that of cooperative learning (Sharan, 1980; Slavin, 1980a). Cooperative learning methods were designed to provide teachers with smallgroup techniques for daily instruction in the classroom. SMALL-GROUP TEACHING The educational approach of small-group teaching defines the class as a group of groups (or an aggregate of groups). The class is organized in groups of two to six students in order to fulfill a learning task cooperatively. The learning task is based on interaction and reciprocal interdependence among the members of the group and requires mutual help. In this educational approach, students and teacher are in a state of dynamic cooperation and together buildup an intimate learning and social atmosphere in the classroom. The textbook and the teacher are no longer the only sources of information but are replaced by a variety of other sourcessuch as books, journals, and consultation with other people. The roles of teacher and students are different from those in the whole-class setup. The teacher is no longer a lecturer or transmitter of material, but rather a facilitator of learning who focuses on the learning process by encouraging cooperation among the students; the students are no longer trying to impress their teacher but are busy learning actively. Small-group teaching is a flexible tool that can be adapted to both the affective and the learning needs of students. It is not one particular teaching technique but rather an overall method which can serve different needs in the language classroom depending on the technique used: Slavins (1978a) Student Teams and Achievement Divisions (STAD), Sharan and Lazarowitzs (1978) Discussion Group (DG), and Aaronson, Blaney, Sikes, and Shapps (1978) Jigsaw. The fundamental principles and operational features of the STAD and DG techniques, which were .implemented in the study, are described below. In the Jigsaw, which is beyond the scope of this article, each student acquires expertise in only one portion of the material given to the whole group. The students who have the same portion form an expert team in which they help each other master their common section. They then return to their own groups, and all teach their expertise to their group mates. When all the students have studied all the portions of the material, they take an individual quiz.

The STAD Technique STAD, a peer-tutoring technique, is based on raising students motivation for learning by focusing on cooperation of members within each team, followed by competition among the teams in the class. The technique follows six stages: (a) organizing small heterogeneous groups, (b) presenting the teaching unit, (c) assigning cooperative peer group work on a worksheet, (d) giving an individual quiz, (e) computing students scores, and (f) announcing the group scores on the bulletin board and rewarding the winning group. The teacher, using whatever information is available (e.g., test scores, grades, or personal judgment), makes a list of the students in the class, ranking them from highest to lowest; these scores serve as the students base scores. The list is divided into four equal parts. The teacher assigns the students to teams of four and balances them so that each team is composed of a representative from each of the four performance-based divisions. Thus, students with different achievement levels within a team can tutor each other, and because the teams are balanced, no single team has an advantage in the competition phase. Students are not informed of their divisional rankings, which are used only as reference groups for the teacher. Every 2-3 months the divisions are reevaluated. Students may move to a different division on the basis of their achievement. The teacher presents the teaching unit in any suitable way. Worksheets are then distributed, and the group members proceed to study together, first in pairs and then in fours. Each pair gets one worksheet and one answer sheet. The items on the worksheets all deal with a specific teaching point such as vocabulary, grammar, or even reading or listening for specific information. All the items are objective; that is, they have only one correct answer. The students study the material together to be sure that all team members learn the material and are prepared for the individual quiz that follows. It is in the interest of the whole team that each student does well. Each students contribution to the team score is figured on the basis of his or her base score. If the quiz score is higher than the base score by 2 points, for example, the students contribution will be 2 points; if the quiz score is equal to the base score or lower, the student will not contribute any points to the team score. No points are taken off for lower achievement. Thus, students compete with themselves (i.e., with their own base scores) and not with their peers. This gives each student an equal opportunity to contribute to the team score. (See Slavin, 1978b, 1980b, for a detailed description of the scoring system. ) The team score is computed and publicized on the class bulletin board, and the winning team is rewarded. This act has strong motivational power.

The STAD technique, as described above, aims at raising student motivation according to Atkinsons (1958) theory of motivation, which posits a multiplicative relationship between the incentive value of success and the probability of success. The establishment of a social game environment via competition between groups offers students an incentive for success, and equal opportunity for success via the special scoring system makes students aware of the probability of success and thus of the fact that everyones additional effort is worthwhile and pays. This does not happen in the traditional grading system, in which high grade marks are hard to acquire by the low-learning-status student and easy to acquire by the high-learning-status student. By this system students motivation is not necessarily raised, since, as is well known, motivation is low when it is either too easy or too difficult to succeed (Atkinson, 1958). By offering each individual an equal opportunity to contribute to the group grade, the STAD technique raises the motivation of all students and prevents a danger that might occur in small-group learning, namely that the contribution of lower level students is not appreciated at all (Slavin, 1983) . The DG Technique The DC technique and its principles are derived largely from the philosophical outlook of John Dewey, further developed by Sharan and Sharan (1975) and Sharan and Lazarowitz (1978) and adapted for EFL needs by Bejarano (1985). The technique follows six stages: (a) choosing the topic of interest and organizing in interest groups, (b) examining the learning task and planning the way to carry it out, (c) using a variety of resources within and outside the school to help carry out the learning task, (d) planning a summing-up group report, (e) reporting the group product to the whole class, and (f) evaluating the overall group product. Effective study in DC necessitates the social skills and functions requisite for group participation, such as attentive listening, effective implementation of peers ideas, cooperation and sharing of information, mutual help, talking in turn, serving as group leader, and so on (Sharan & Sharan, 1975, p. 65). All of these are basic skills for interpersonal and multilateral communication. The EFL teachers role is to set up tasks that facilitate the acquisition of these skills and that at the same time serve a language teaching goal. The following activity is an example of a task, the purpose of which is to practice reported speech and to train for attentive listening.

1. The class is divided into groups of four. 2. The teacher sets up the task by explaining that each student should tell his or her group peers about things the student likes or dislikes and about a personality the student respects and why. Each speaker must report what the preceding speaker said before beginning his or her own talk. Inaccurate reports are corrected by the group. 3. The teacher allows the activity to continue for about 10 minutes. Each group then discusses how well its members listened to each other. 4. The entire class discusses the exercise. The following questions may be used as a guide: Was it difficult to report your peers talk? Was the difficulty due to the lack of clarity in the statement itself, to the length of the statement, or to the number of ideas expressed in it? 5. The class discusses ways to improve attentive listening and consequently more efficient communication among the students. Learning tasks suitable for the DG technique are usually based on topics for discussion or problem-solving issues which require bilateral or multilateral communication, negotiation, and interaction among participants. It is the structure of the learning task which stimulates students active participation in mutual communication. The task is divided into subtasks, with each participant responsible only for his or her share (Sharan & Lazarowitz, 1978). The task creates mutual dependence among the group members, since it cannot be completed without each individuals contribution. This results in obligatory and instrumental multilateral communication within the group (Barnes, 1976; Johnson & Johnson, 1975; Joyce& Weil, 1972; Sharan & Lazarowitz, 1978) and is an effective cohesive element both socially and in terms of the need to carry out the group task (Pepitone, 1980; Schmuck& Schmuck, 1979). The learning task can be of different kinds: 1. Topics for research: These are divided into subtopics, each of which is investigated thoroughly by an individual or a pair of students. A variety of resources (e.g., personal interviews, books, articles, radio and television programs, etc. ) is used. All members then share and combine their individual products and synthesize them into a group product. 2. Topics for discussion: These are based on information gaps. Each individual holds different pieces of information, and the task cannot be completed until all members of the group have

exchanged and shared all of their pieces of information. This stimulates active participation in the discussion. Such activities include problem solving, simulation games, and role playing. The differences and similarities between the three methods used in the experiment and their relevance to EFL are presented in the Appendix. THE STUDY Although the use of group work in the EFL classroom has been recommended in the last few years and teaching materials have been developed accordingly, little or no attention has been given to describing the roles of the teacher and the students; the social skills required for participating in small-group tasks; and the planning, structuring, and management of the learning tasks, which should be based on the underlying principles of the techniques as they are practiced in the classroom (see Long & Porter, 1985, p. 224). The effect of carefully outlined group methods on the achievement of EFL learners has received little attention from researchers. This is also true of the relationship between group work and wholeclass, teacher-led instruction (see Long & Porter, 1985, p. 225). The purpose of this study was (a) to assess the effect of two cooperative small-group learning methods (DG and STAD) and of the traditional whole-class method on the general achievement of junior high school EFL learners and on their acquisition of specific language skills and (b) to check whether there is a differential effect of small-group methods on the acquisition of different language skills. Subjects Teachers and Pupils A total of 781 students studying English in 33 seventh-grade classes in three junior high schools participated in the experiment (see Table 1). The students, who were in their third year of English in school, represented average levels of achievement in Israel. The experiment was conducted during five school periods a week, for 4 months. Of the 781 students, 665 were present for both the preand posttests. Only these 665 cases with complete data were included in the statistical analysis. The 18 participating teachers, who represented the entire Englishteaching staff at the three schools, were assigned to methods and classes at random, most of them teaching more than one class. All teachers participated in a very intensive preexperiment training

TABLE 1 Number of Students and Classes in Each of the Three Teaching Situations Wholeclass




program. This consisted of 12 workshops, each lasting 2 hours, led by experts in EFL and in the specific teaching methods to be used in the study. The teachers were trained both for the specific skills required to utilize the method to which they were assigned and for the immediate implementation of that method in the EFL classroom. Under the guidance of the researcher, the teachers developed specialized activities and materials based on the principles underlying those teaching methods. During the experiment, the researcher and a colleague continued to visit the classrooms for in-class follow-up and coaching. Teachers assigned to the whole-class method also went through a training period, despite their claim that they had nothing to change. The purpose was to ensure that teachers skills in this group were sharpened and that their classes could serve as enriched control groups. The Achievement Test A specially constructed achievement test was given twice, as both pre- and posttest. Thus, we could evaluate certain areas of language functioning and note the changes that occurred during the experimental period. The test included three components: two integrative tests (listening comprehension and reading comprehension) and a discrete-point grammar and vocabulary test (see Oller, 1979, p. 25). The listening comprehension test assessed (a) comprehension of recorded statements describing the details in pictures in the test book, (b) the ability to carry out recorded instructions by performing tasks on given pictures, (c) understanding of a recorded informative text, and (d) understanding of a recorded narrative text. The reading comprehension test assessed (a) the ability to read a sentence and match its content to a given picture, (b) reading and

understanding of an informative text, and (c) reading and understanding of a narrative text. The aim of the discrete-point grammar and vocabulary test was to assess (a) proper use of basic grammatical structures and question formation (with which Hebrew speakers experience particular difficulty) and (b) appropriate use of vocabulary items. Classroom Observations To gather systematic data on the implementation of the three teaching methods, three trained observers visited every class twice during the experimental period. (These observations were completely independent of the follow-up and coaching visits. ) Interrater reliability, after a period of training with videotapes and classrooms not participating in the project, was 85% on two separate trials. Each observation was carried out during classroom sessions when no tests or quizzes were in progress. Raters used a 20-item observation schedule adapted from the method evolved by Thew (1975). Ratings were recorded every 15 seconds for a 7-minute period at the beginning, middle, and end of a classroom session. These observations were studied to determine whether STAD or DG teachers were implementing the assigned methods. Two teachers were found to have failed to implement the DG techniques; data gathered in their classes were included in the whole-class data. (This change is reflected in the figures presented in Table 1.) The observers were also asked to write an impressionistic report of whatever they could detect in the classes concerning students behaviors while they were carrying out the various activities. ANALYSES AND RESULTS Table 2 presents pre- and posttest means of each of the three student groups for each of the scales of the achievement test separately (listening comprehension, reading comprehension, and the discrete-point scale) as well as for the scores from the test as a whole (total). An analysis of variance (ANOVA) revealed that there was a significant difference in the degree of improvement from preto posttest generated by the different methods on two out of the four scales: the total score, F (2, 662) = 3.75, p <.05, and the listening comprehension scale, F = 6.88, p <.005. No significant differences were found between the three methods on either the reading comprehension scale or the discrete-point scale.

TABLE 2 Descriptive Statistics for Pre- and Posttest Performance of Students in the Three Teaching Situations


(n = 29) Scale (range of scores) Pre

Total M SD Listening comprehension (0-57) M SD Reading comprehension (0-19) M SD Discrete-point (0-15) M SD 50.52 19.59

(n= 198) Pre

53.94 19.94

Whole-class (n= 238) Pre

57.58 17.49

57.28 20.87

61.39 21.35

62.45 18.31

31.51 11.78

36.00 12.91

34.46 11.83

38.80 13.02

36.74 10.47

38.97 11.15

10.97 5.43 8.04 4.30

12.48 5.36 8.80 4.35

11.63 5.19 7.85 4.49

13.39 5.37 9.20 4.54

12.23 4.10 8.61 4.48

14.10 4.70 9.38 4.28

To determine the relative extent of the change fostered by the various instructional methods, three 2 x 2 ANOVAs were carried out, with methods (2) the between-group variable and time (repeated measures of pre- and posttest) a within-group variable. The three comparisons were DG versus whole-class, STAD versus whole-class, and DG versus STAD. These analyses were carried out on the total test score, on the listening comprehension score, and on the discrete-point score, with the following results: 1. Pupils in the DG classes registered greater improvement than their peers in the whole-class setting on both the total test, F (l, 465) = 4.23, p <.05, and on the listening comprehension scale, F (1, 465) = 11.99, p <.001. 2. Pupils in the STAD classes also achieved higher scores than those in the whole-class setting on the total test, F (l, 434) = 6.27, p <.05, and on the listening comprehension scale, F (l, 434) = 8.60, p <.005. 3. Pupils in the STAD classes achieved higher scores than did pupils in the DG classes on the discrete-point test, with the difference approaching the significance level, F (l, 425) = 3.33, p = .06. 492 TESOL QUARTERLY

4. No significant difference was found on the achievement scores of pupils in the DG classes compared with those of pupils in the STAD classes on the total test score or on the listening comprehension score. Thus, both small-group methods proved superior to the whole-class method on these scales, and the two group methods emerged as equally effective. To summarize, on the listening comprehension scale, the size of the change that occurred in the DG and STAD classes from pretest to posttest (4.49 and 4.34, respectively) was bigger than that in the whole-class setting (2.23). The change that occurred on the discretepoint scale was bigger in the STAD classes (1.35) than in the DG classes (0.76) and in the whole-class setting (0.77). These findings illustrate the greater effectiveness of the two group methods in developing listening comprehension skills and the greater effectiveness of the STAD technique compared with the DG and whole-class methods in teaching and learning discrete points (such as grammar and vocabulary) in the EFL classroom. DISCUSSION The Effect of Small-Group Teaching on General Achievement What promoted general higher language achievement (the total test score) in the classes utilizing small-group techniques was apparently the active communication approach (see Figure 1) used in this experiment; the processing of both the instructional content and the social and communicative skills, which occurs within the experiential context of the small-group setting, was enhanced by means of appropriate learning tasks. Indeed, these social processes encouraged the active participation of all members of the group and intensive interaction among them. This interaction was clearly discerned by the observers, who reported that the students were actively involved in real communication rather than in using answers taken out of texts or manipulating given information or set linguistic structures. This real, or instrumental, communication was a result of an immediate need for interpersonal contact. Since the communication activities in which they were involved were meaningful, the students were motivated to sustain that communication in order to accomplish the group task. This is in line with Taylors (1983, p. 72) view of the features classroom instruction should incorporate. It is also in accordance with the communicative approach to language teaching, which assumes that language acquisition occurs with intensive engagement in extended discourse in real communicative contexts (Breen & Candlin, 1980; Krashen, 1981; Morrow, 1981; Taylor, 1983; Widdowson, 1978).

FIGURE 1 Active Communication Approach

I Processing of instructiomd content 1. Structural component 2. Functional component 3. Communicative component

Appropriate learning tasks which facilitate I and II within experiential context


Academic achievement t

II processing of sociaI and communicative skik in small group

This approach actually dictated the structuring of the learning tasks by taking into consideration the linguistic content to be taught (what) and the process of teaching the linguistic content (how). The learning tasks provided for the incorporation of the three linguistic components (structural, functional, and communicative), as well as for the processing of social factors in the small-group setting, an appropriate social and learning atmosphere. Extended linguistic practice was thus facilitated in a cooperative setting, in which the roles of teacher and students differed significantly from those in the whole-class setting. The fulfillment of the learning task, entailing a successful bridging of information gaps, depended on the active participation of each student in the group, in an atmosphere of interdependence, cooperation, and mutual help (Sharan & Lazarowitz, 1978). The 494 TESOL QUARTERLY

tasks, constructed so as to fulfill the linguistic aim, on the one hand, and to suit the specific features of the group method, on the other, included (a) practice of discrete linguistic points (such as grammar and vocabulary) via STAD, whose specific features facilitate the teaching of these basic skills, and (b) involvement in global communication, facilitated by using the DG technique in such tasks as role playing, simulation, problem solving, and so on. The observation carried out in all classes indicated that all students were involved in interaction and communication among themselves during at least 40% (and sometimes up to 80%) of class time, whereas in the traditional classes the teacher lectured during at least 80% of class time. The extended active use of the language provided learners with opportunities to practice linguistic structures taught formally in the classroom and also to acquire structures not taught formally. During group interaction, the learners apparently used new linguistic structures necessary for specific task communication, and these enriched their linguistic competence. Being involved in meaningful discussion, they probably invested intellectual resources to receive and transmit messages, despite difficulties in expressing themselves in the new language. The learning tasks in the present study indeed fostered language acquisition as well as language learning by setting up an immediate need for the active participation of every individual in the group. This agrees with Hatchs (1978) finding that linguistic forms are acquired and used productively only when it assumes a critical role in transmitting essential information (p. 472). In addition to the learning task, the change in the roles of teacher and students apparently also contributed to the positive results obtained in this study. The teachers role changed from that of criticizing authority to that of adviser and facilitator. This provided ample opportunity for mutual, unmediated, and spontaneous communication ( Widdowsons aspect of use). In the whole-class setting, teachers spent a considerable amount of time asking questions which pupils were expected to answer in linguistically accurate formin what Barnes (1976) calls final draft formulationthus limiting spontaneous and uninhibited language use. In contrast, peer discourse in the small-group setting apparently permitted pupils to think aloud and, by necessity, to talk in less polished language. There was far less inhibition and tension under these circumstances because discourse served communicative needs rather than the demands of public recitation. The opportunity for mutual communication exchange among peers, despite the danger of students hearing and producing many errors, is at least as important for language acquisition as the errorA COOPERATIVE SMALL-GROUP METHODOLOGY 495

free lesson presented by the teacher in the traditional setting. Peer communication is also more likely to relate the content of the verbal message to the pupils experience outside the domain of the linguistic forms per se. It is widely acknowledged that second language learning should focus on relations between language and experiential contexts which provide meanings to words, an argument proposed by John Dewey decades ago. Furthermore, peers in small groups are more likely to employ language comprehensible to each other than is the teacher and, when the need arises, to ask each other to clarify meanings. On all these accounts, small-group discussion appears to be a potentially more meaningful social environment for promoting language use and comprehension than the traditional classroom. The precise manner in which the process of language learning unfolds in the small-group setting deserves intensive research, which only microanalytic studies of peer interactions in groups can document. The Effect of Small-Group Teaching on Listening Comprehension and Discrete Points The finding that the group techniques attained superior results on the listening comprehension scale seems to be due to the fact that the learning tasks (especially in the DG classes) required verbal communication which involved speaking and listening intermittently rather than reciting as called upon by the teacher in the traditional classroom. This kind of communication, in which every individual had to participate actively, required instrumental listening and attention in order to react in time. Thus, students learning in small groups spent more time in active, instrumental listening than did students in the whole-class setting, where the teacher very often speaks over the students heads. Krashen (1981) also found that students prefer to listen to their friends rather than to the teacher, and Cohen (1984) reported that in the whole-class lesson only 2-5%50% of the students actually listen to the teacher. The positive social environment on which the group techniques are based must also have contributed to these results. It is very possible that the speaking ability of the students improved together with their listening ability, but this was not tested in the present study. The findings that the two group methods promoted significantly superior achievement compared with the whole-class method on the listening comprehension scale and that the STAD technique promoted higher achievement on the discrete-point scale serve as

partial support for the hypothesis that different group techniques serve different instructional objectives in the language classroom. The fact that no differences were found between the two group techniques on the listeniug comprehension scale, despite the basic difference between the two techniquesSTAD emphasizing learning of structural language material presented initially by the teacher and DG emphasizing global use of language in real communicationrequires explanation. The observations carried out in the classes indicated that although the STAD teachers were instructed not to allow multilateral communication within the groups but to ensure concentration on peer tutoring of previously taught material, such group conversations did develop, and the teachers did not want to interrupt them. This probably had its effect and may explain the similar results obtained for both group methods on the listening comprehension scale. The Effect of Small-Group Teaching on Reading Comprehension The fact that no significant differences between the small-group and the whole-class methods were observed on the reading comprehension scale also requires attention. It seems that the specific tasks constructed on the principles of the small-group teaching techniques employed in this study did not promote the acquisition and practice of reading strategies. Moreover, the students in the small-group classes were neither given tasks designed to develop reading strategies, nor were they encouraged to do so in any way. The Jigsaw technique, not used in this study, might prove to be effective in the acquisition of reading comprehension skills. SUMMARY The results obtained here, in which the STAD technique promoted higher results on the discrete-point scale, indicate that STAD supplements and complements the DG technique, which, despite the statistical findings, seemed to be more efficient for practicing global language skills. These results answer the basic requirements of the modern communicative approach to language teaching: basic knowledge of grammar and vocabulary (Widdowsons usage and Littlewoods grammatical component), on the one hand, and functional competence, which entails the ability to implement the linguistic knowledge in natural, authentic communicative situations (Widdowsons use and Littlewoods functional/communicative component), on the other (see Figure 2).

FIGURE 2 A Cooperative Small-Group Methodology in the Language Class

Appropriate learning tasks

Based on the theoretical principles underlying this research and the findings reported here, a cooperative small-group methodology in the language classroom is recommended. The different group techniques complement one another; they serve different teaching objectives in the language class and thus form the link between the teaching content (what) and the teaching process (how). Implementation of this approach requires intensive teacher training for use of the techniques, both in terms of operational procedures in the classroom and in terms of appropriate design of the learning tasks.



ACKNOWLEDGMENTS The study reported here is based on a PhD dissertation carried out as part of a larger project directed by Professor Shlomo Sharan (School of Education, Tel-Aviv University). Dr. E. Olshtain served as a consultant for the study. THE AUTHOR Yael Bejarano, PhD, is the head of the EFL Department at the Open University of Israel, where she administers language programs and is involved in teacher training and materials development. Her research focuses on small-group teaching and second language testing. She is currently serving as the ISRATESOL newsletter editor. REFERENCES
Aaronson, E., Blaney, N., Sikes, J., & Shapp, M. (1978). The jigsaw classroom. Beverly Hills, CA: Sage. Asher, J.J. (1977a). Children learning another language: A developmental hypothesis. Child Development, 48, 1040-1048. Asher, J.J. (1977b). Learning another language through actions: The complete teachers guidebook. Los Gates, CA: Sky Oaks Productions. Atkinson, J. W. (1958). Towards experimental analysis of human motivation in terms of motives, expectancies and incentives. In J. W. Atkinson (Ed.), Motives in fantasy, action and society (pp. 288-305). Princeton, NJ: Van Nostrand. Barnes, D. (1976). From communication to curriculum. Middlesex, England: Penguin Books. Bejarano, Y. (1985). The effects of teaching English as a foreign language in small cooperative groups on the achievement of pupils in the multiethnic classroom. Unpublished doctoral dissertation, Tel-Aviv University. Breen, M. P., & Candlin, C.N. (1980). The essentials of a communicative curriculum. Applied Linguistics, 1, 89-112. Brumfit, C. (1980). Problems and principles in English teaching. N e w York: Pergamon Press. Brumfit, C., & Johnson, K. (1979). The communicative approach to language teaching. London: Oxford University Press. Cohen, A. (1984, March). Introspecting about second language learning. Paper presented at the Ninth ILASH Conference, Netanya, Israel. Curran, C.A. (1972). Counseling learning: A whole person model for education. New York: Grune and Stratton. Curran, C.A. (1976). Counseling learning in a second language. Apple River, IL: Apple River Press. Gattegno, C. (1972). Teaching foreign languages in schools: The silent wag. New York: Educational Solutions. Gattegno, C. (1976). The common sense of teaching foreign languages. New York: Educational Solutions.

Hatch, E. (1978). Second language acquisition. Rowley, MA: Newbury House. Hymes, D. (1972). On communicative competence. In J.B. Pride & J. Holmes (Eds.), Sociolinguistics: Selected readings (pp. 269-293). Harmondsworth, England: Penguin Books. Johnson, K., & Johnson, R. (1975). Learning together and alone. Englewood Cliffs, NJ: Prentice-Hall. Johnson, K., & Morrow, K. (Eds.). (1981). Communication in the classroom. London: Longman Group. Joyce, B., & Weil, M. (1972). Models of teaching. Englewood Cliffs, NJ: Prentice-Hall. Krahnke, K.J., & Christison M.A. (1983). Recent language research and some language teaching principles. TESOL Quarterly, 17, 625-649. Krashen, S. (1981). Second language acquisition and second language learning. New York: Pergamon Press. Littlewood, W. (1981). Communicative language teacher: An introduction. New York: Cambridge University Press. Long, M. H., & Porter, P.A. (1985). Group work, interlanguage talk, and second language acquisition. TESOL Quarterly, 19, 207-227. Lozanov, G. (1979). Suggestology and outlines of suggestopedy. N e w York: Gordon and Breach. Morrow, K. (1981). Principles of communicative methodology. In K. Johnson & K. Morrow (Eds.), Communication in the classroom (pp. 5970). London: Longman Group. Oller, J. (1979). Language tests at school. London: Longman Group. Pepitone, E. (1980). Children in cooperation and competition. Lexington, MA: Lexington Books. Schmuck, R., & Schmuck, P. (1979). Group processes in the classroom (3rd cd.). Dubuque, IA: Brown. Sharan, S. (1980). Cooperative learning in small groups: Recent methods and effects on achievement, attitudes and ethnic relations. Review of Educational Research, 50, 241-271. Sharan, S., & Lazarowitz, R. (1978). Cooperation and communication in school [in Hebrew]. Tel-Aviv and Jerusalem: Shocken. Sharan, S., & Sharan, Y. (1975). Small group teaching [in Hebrew]. TelAviv: Shocken. Slavin, R.E. (1978a). Student teams and achievement divisions. Journal of Research and Development in Education, 12, 39-49. Slavin, R.E. (1978b). Using student team learning (The Johns Hopkins team learning project). Baltimore: The Johns Hopkins University. Slavin, R.E. (1980a). Effect of student teams and peer tutoring on academic achievement and time on task. Journal of Experimental Education, 48, 252-257. Slavin, R.E. (1980b). Student team learning A manual for teachers. In S. Sharan, D. Hare, C. Webb, & R. Hertz-Lazarowitz (Eds.), Cooperation in education (pp. 82-135). Provo, UT: Brigham Young University Press. Slavin, R.E. (1983). Cooperative learning. New York: Longman Group. 500 TESOL QUARTERLY

Stevick, E.W. (1980). Teaching languages: A way and ways. Rowley, MA: Newbury House. Taylor, B.P. (1983). Teaching ESL: Incorporating a communicative, student-centered component. TESOL Quarterly, 17, 69-88. Thew, D. (1975). The classroom social organization category system. The Classroom Interaction Newsletter, 2, 18-24. Widdowson, H.G. (1978). Teaching language as communication. London: Oxford University Press. Winitz, H., & Reeds, J. (1973). Rapid acquisition of a foreign language (German) by avoidance of speaking. International Review of Applied Linguistics, 2, 295-317.



APPENDIX Differences and Similarities Between DG, STAD, and Whole-Class Methods and Their Relevance to EFL

Instructional method Element compared

Source and variety of information

Information gathered by the pupils; learning sources varied in number and kind. Tasks stress problem solving, interpretation, synthesis, and application of information. Task is subdivided; cannot be completed unless all individual contributions to final group product are made.



Relevance of small-group methods to EFL

. In DG method students relative control of information sources provides motivation. Both group methods encourage active participation through mutual exchange based on information gaps or peer tutoring and drilling. Both group methods necessitate active participation.

The teaching task . Information transmitted Information transmitted by the teacher or a text; by the teacher or a text. learning sources limited to cards, worksheets, or lecture. Tasks emphasize acquisition of information or basic language skills. Tasks stress passive understanding and internalizing teachertaught or text material. All students are assigned identical tasks; individuals pursue their own success.

Nature of learning tasks

Completion of learning task

Each individual is interested in the success of group peers in final quiz; results add up to the group score.

Group goal

Group members cooperate and coordinate activities on a group-wide basis to achieve the common goal: completing the learning task. Both receptive and productive skills are used simultaneously.

Group members cooperate to achieve the common goal: a high group score in order to compete with the other groups.

There is no common group goal.

In both group methods the common group goal serves as a means of motivation.

Use of language skills

Both receptive and productive skills are used simultaneously but in a more controlled manner than in DG.

Language skills are used in an integrative manner.

In both group methods there is active and integrative use of both receptive and productive skills.

Kind of communication

Peer communication within teams: bilateral or multilateral.

Interpersonal communication and relations Primarily unilateral; Peer communication sometimes develops into within teams primarily unilateral and bilateral: bilateral; sometimes (student <> teacher) multilateral. though these interactions are usually very short. Rehearsal of teachertaught material or filling in worksheets and quizzes. Primarily for clarification of teaching points.

Bilateral and multilateral communication in both group methods.

Goal of communication

Mutual exchange for problem clarification, interpretation, and exchange of ideas.

Intensive and active use of the new language in both group methods, especially in DG.

APPENDIX (Continued) Differences and Similarities Between DG, STAD, and Whole-Class Methods and Their Relevance to EFL Instructional method Element compared
Extent of communication

Free and global use of the language especially in topics for discussion. Unmediated interaction among members of the small group, usually based on cooperation and mutual help. Positive.

More limited use of the language compared with DG, though used freely in connection with the specific material taught. Unmediated interaction among members of the small group, usually based on cooperation and mutual help. Positive.

Almost impossible for every student to participate actively in whole-class discussion. Almost no interaction among students; when occurring, it is in front of the teacher and classmates. Individuals pursue own learning tasks and vie for personal status with teacher. Main source of knowledge and the high authority in the system.

Relevance of small-group methods to EFL

Intensive and active use of the new language in both group methods, especially in DG. Cooperation and mutual help in a positive social atmosphere in both group methods. Positive social atmosphere in both group methods.

Interpersonal interaction

Interpersonal relations

Role of the teacher

Guide, consultant, and facilitator.

Major source of knowledge but encourages interpersonal communication, interaction, and mutual help.

In both group methods the teacher is no longer the only authority, but serves as a consultant and facilitator.

TESOL QUARTERLY, Vol. 21, No. 3, September 1987

English Language Proficiency and the Prediction of Academic Success

JANET G. GRAHAM University of Maryland, Baltimore County

This article discusses the relationship between English language proficiency and academic success in universities and colleges in which English is the language of instruction. It points out some of the difficulties associated with determining this relationship and summarizes previous investigations of the issue. It is argued that while the research clearly shows that many factors other than English proficiency are important to academic success, there may be for each institution, or even for each program within an institution, a minimum level below which lack of sufficient proficiency in English contributes significantly to lack of academic success. Such a level can be determined by each institution individually, but until it is determined, a number of steps can be taken for establishing reasonable English language proficiency requirements.

At many colleges and universities, ESL professionals are called upon by admissions officers for help in making admissions decisions for nonnative speakers of English. ESL professionals may be asked what cutoff scores should be used on standardized tests that have already been administered, such as the Test of English as a Foreign Language (TOEFL) or the Michigan Test of English Language Proficiency (MTELP). Or they may be asked to test the English language of applicants who do not already have standardized test scores and, on the basis of these tests, to make judgments about the adequacy of the applicants English. In both these situations, two assumptions are being made. The first is that a certain level of proficiency is necessary for successful college-level work where English is the language of instruction. The second is that an ESL professional can be expected to know what that level is. A conscientious ESL professional might look to the literature on language proficiency and academic performance for

help in determining appropriate levels. Unfortunately, a search of this literature will reveal that the relationship between English language proficiency and academic success is murky indeed. This article considers some of the problems associated with interpreting research on the relationship between English language proficiency and academic success. It provides a brief review of some of the research of this type, discusses the implications of the research, and offers some suggestions for ways to advise admissions offices in regard to nonnative English-speaking applicants. PROBLEMS WITH ACADEMIC PREDICTION STUDIES Most research into the relationship between English language proficiency and academic success falls under the rubric of academic prediction studies. Such studies investigate relationships between various kinds of informationsuch as test scores, personality factors, previous academic ratingsand academic success. Typically, prediction studies of English language proficiency and academic success calculate correlation coefficients between English proficiency test scores and first-semester grade point average (GPA). There are a number of problems with both the design and interpretation of such studies, including (a) the criterion for judging academic success, (b) the validity of measures of English proficiency, (c) the interpretation of any relationships found, and (d) the large number of uncontrolled variables involved in academic success. While GPA is the most commonly used criterion for academic success, some researchers have noted that it is not always a valid indicator of academic achievement. Heil and Aleamoni (1974), for example, point out that GPA does not take into account the number of courses taken. Students may be able to handle only two courses at a time, due to poor English proficiency, for example, but their GPAs would not reflect this. Heil and Aleamoni also allude to the widely recognized problem of teachers giving sympathy or goodwill grades to nonnative speakers. For studies of graduate students, an additional problem is the limited spread of grades typically given in graduate schools, which means that significant correlations are less likely to be found. Ho and Spinks (1985) argue that GPAs are composed of heterogeneous or divergent elements, especially at the university level, where various academic subjects demand divergent competencies or dispositions (p. 258). For example, some students might have a gift for logical argument, which would serve them well in one course, and a deficiency of

math skills, which would doom them in another. In defense of the use of GPA, it should be pointed out that a study of 2,075 foreign students at the University of California, Los Angeles (UCLA), found that first-semester GPA was the best index of the students eventual success (Sugimoto, 1966). Some prediction studies use other measures of academic success (as can be seen in the brief descriptions in the following section). This variability of the criterion for academic success must be borne in mind when the results of one study are compared with those of another. Another problem with prediction studies has to do with the definition of English language proficiency. Usually, scores on various commercial tests, such as the TOEFL, the MTELP, or the Comprehensive English Language Test (CELT), are used as the measure of proficiency, and proficiency is defined by performance on the test. Therefore, the definition of proficiency is somewhat different in a study using TOEFL scores from that in a study using CELT scores. This may not be as big a problem as it first appears, however. Many studies have found very high correlations between the various well-known tests (J.B. Carroll, 1972; Cervenka, 1978; Dizney, 1965). But with the recent attention to communicative competence (as opposed to linguistic knowledge), the value of traditional multiplechoice proficiency tests, such as the ones named above, for predicting the ability to use the language competently has been severely questioned. B.J. Carroll (1980), for example, argues that when the aim is to measure the ability to use the language (as opposed to knowing the proper usage), tests should not be based on a selection of items chosen on linguistic grounds alone (p. 8). Dulay, Burt, and Krashen (1982) lend support to this argument. As rationally appealing as this argument is, however, it has yet to be conclusively demonstrated that traditional tests do not predict language performance. To the contrary, a number of studies have shown high correlations between traditional multiple-choice tests and integrative tests (such as cloze tests) and performance tests (such as a writing test) (see Hanania & Shikhani, 1986). At least one study (Farhady, 1983) showed fairly strong relationships between an especially designed functional test for university students and various more traditional tests, including tests of grammar. While a number of studies do show very strong correlations between various measures of English language proficiency, the fact that the prediction studies reviewed in the following section use different measures must be remembered when attempting to interpret the results.

Another problem in interpreting results is determining the real significance of a relationship (as expressed by a Pearson productmoment correlation coefficient, for example) as opposed to its statistical significance. A statistically significant relationship does not necessarily indicate a strong relationship, and in fact different researchers will use different cutoff points for claiming strong and weak correlations. Rather than accepting a researchers assessment of the strength of a relationship, it is more useful to interpret a correlation coefficient by squaring it to obtain the common variance between two measures. A correlation coefficient of .40 between TOEFL scores and GPA, for example, explains only 16% of the variance (.42 = .16); 84% of the variance is not accounted for by the TOEFL score. (See Hatch& Farhady, 1982, pp. 201-203, for a more complete explanation of common variance.) Other factors may influence the results of prediction studies, including the number of subjects, their countries of origin, their native language, whether or not they are visiting foreign students or permanent residents or refugees, whether they are undergraduate or graduate students, their majors or areas of specialization, and the design of the study. Not only do variables such as these affect the outcomes of the studies, they limit the generalizability of the studies and make direct comparisons difficult. A REVIEW OF SOME PREDICTION STUDIES The following summaries give an idea of the variety of academic prediction studies that have been done and the difficulty of using their findings to generalize about the relationship of English proficiency to academic success. Because the studies are not directly comparable, they have been grouped not according to numerical criteria such as correlation coefficients, but according to the conclusions drawn by the researchers, which in some cases may differ from the conclusions readers would draw from the same data. The studies have been grouped in the interest of clarity, not to advance any particular argument. Negative Conclusions A number of researchers have concluded that there is little relationship between English language proficiency and academic success. Mulligans (1966) study of 669 students at the School of Business and Public Administration of the City College of New York produced results which ran counter to the researchers assumption that there would be a strong correlation between

English proficiency and college performance (p. 310). These students had all been educated abroad but were immigrants or refugees rather than visiting foreign students. English proficiency was defined by placement in the freshman remedial English class or in the regular freshman English class. Academic success was defined as performing above a C average. There was no significant relationship between English placement and a GPA of C or above, leading Mulligan to conclude that deficiency in English was not a significant deterrent to scholastic achievement (p. 313). Similar results were obtained in a study of 2,075 foreign students at UCLA in 1964-1965 which ran correlations between a number of different variables and academic success (Sugimoto, 1966). Academic success in this study was defined in terms of whether or not the student had obtained a degree, certificate, or credential or had obtained permission to continue university studies at a higher level. English language proficiency, as measured by the institutions own tests, was one of the least predictive of the variables in the study (r = .046), leading the researcher to recommend giving less emphasis to English examination scores in the making of admissions decisions. Hwang and Dizney (1970) also found that English language test scores were poor predictors of academic performance. They found no significant correlation between TOEFL scores and the first-term GPA for 63 Chinese graduate students at the University of Oregon (r = .19). It should be noted that most of these students were specializing in areas requiring significant use of English: 21 in education, 16 in social sciences, and 6 in architecture. Interestingly, Hwang and Dizney found a correlation of .66 between ESL course grades and overall GPA, but the course grades almost certainly measured academic skills and nonacademic factors as well as English proficiency. Sharon (1972) attempted to determine whether the TOEFL added to the predictive value of the Verbal Ability section of the Graduate Record Exam (GRE-V). The subjects were 975 foreign graduate students from 24 schools. A relatively high correlation ( r= .70) was found between the GRE-V and the TOEFL, but not between the TOEFL and GPA (r = .26) or between the TOEFL combined with the GRE-V and GPA (r = .27). The Graduate Record Exam of Quantitative Ability (GRE-Q) turned out to be the best single predictor (r = .32); combining TOEFL with GRE-Q raised the correlation to .34. According to Sharon, it appears that foreign students with low English verbal aptitude can succeed in American graduate schools (p. 431). ENGLISH PROFICIENCY AND ACADEMIC SUCCESS 509

Another study of the TOEFL and GRE-V as predictors was conducted by Shay (1975), who also found that these tests failed to predict academic performance. The correlations between TOEFL and GPA ranged from .08 to .12 and between GRE-V and GPA from .06 to .09; none of the correlations was statistically significant. A study of foreign undergraduates by Wilcox (1975) revealed significant correlations between high school GPA and academic success (about .50), math aptitude and academic success (about .50), and the combination of math aptitude, high school GPA, and academic success (about .60) but found that correlations between TOEFL and GPA were considerably different for the two groups of students in the sample. For the group from Hong Kong, the correlation was .00; for the group from Vietnam, however, the correlation was .46. Wilcoxs conclusion was that overall, neither verbal aptitude nor English proficiency contributed to prediction (p. 6084B). Gue and Holdaway (1973) reached a similar conclusion. They found statistically significant but not strong correlations between TOEFL and GPA for their 123 Thai education majors. These students were tested both before and after a summer language program. The correlation between the summer TOEFL scores and GPA was .49 and between the fall TOEFL scores and GPA was .59, both significant at the .01 level. While some researchers might label these correlations rather strong, Gue and Holdaway concluded that English proficiency was simply not a good predictor because a number of other factors, such as motivation and homesickness, may combine to offset good language proficiency, or to overcome initially lower proficiency (p. 102). Mixed Conclusions A number of studies of the relationship between English language proficiency and academic success have led the researchers to mixed or qualified conclusions. Slark and Bateman (1982) used the NelsonDenny Reading Test and the Listening section of the CELT to measure the English proficiency of nonnative undergraduates. Vietnamese was the most common language of the subjects, followed by Spanish. Class grades, rather than GPA, were used as the criterion for academic success. The Nelson-Denny total scores correlated significantly with success in 4 out of 22 courses, while Vocabulary subscores and Comprehension subscores correlated significantly with course grades in 6 out of 22 and 8 out of 22 courses, respectively (the significant Spearman rho rank-order

correlation coefficients ranged from .388 to .590). CELT Listening scores correlated significantly with grades in 9 out of 22 courses. With a sample of 154 freshman foreign students, a study of the predictive value of the TOEFL as it is used in Oklahoma colleges and universities for freshman admissions (Bostic, 1981) found significant but not large positive correlations (r = .169) between TOEFL scores and overall GPA. An interesting wrinkle in this study was that the researcher looked at the relationship between the TOEFL scores and both language oriented fields and scientifically oriented fields and found significant positive correlations with the scientifically oriented field GPA (r = .50) but not with the language-oriented field GPA (r = .08). Mestre (1981), who investigated the relationship of a number of mathematics and language variables to academic success among 60 bilingual Hispanic students, found significant correlations (ranging from .30 to .48) between all sections of a reading test (Manuel, 1962) and GPA. However, the Verbal Ability section of the Scholastic Aptitude Test (SAT-V) did not correlate significantly with GPA (r = .12) among the bilingual students, even though it did with monolingual students in the study (r = .35). This finding led Mestre to question the validity of using the SAT-V as a basis for admissions decisions for bilingual students. TOEFL scores and grades in a preuniversity English program were used as the measures of English proficiency in a study of 159 undergraduates and graduates at the University of Arizona (Stover, 1982). Conflicting outcomes were found: Both undergraduates and graduates with TOEFL scores of less than 500 were able to achieve at an acceptable level in their first semester. However, while the TOEFL scores and the GPAs in the preuniversity English program were significantly related to academic success in the case of the undergraduates (r = .21, p = .05), they were not significant in the case of the graduate students (r = .13). A recent study of the value of the TOEFL score as a predictor of academic success for 376 graduate students at the State University of New York at Albany (Light, Xu, & Mossop, 1987) found a statistically significant but weak correlation (r = .14) between TOEFL scores and GPA. This study is particularly interesting because of two other significant findings: (a) The relationship between TOEFL and CPA varied considerably when students were grouped by major area of study, and (b) the higher the TOEFL score, the more graduate credits tended to be earned during the first semester.

Positive Conclusions About the same number of researchers appear to have concluded that English language proficiency is a useful predictor of academic success as have not. Burgess and Greis (1970), using a sample of 17 college students, found that TOEFL did correlate significantly with grade point average, particularly when grades for courses requiring little English (such as art, music, and math) were deleted from the grades being averaged (TOEFL with total GPA, r = .53; TOEFL with weighted GPA, r = .56). Writing was also found to be a good predictor of the total GPA, r = .64; with weighted GPA, r = .66), while listening ability was not (.30 and .57). They concluded that proficiency in reading and writing English was important to college success. Heil and Aleamoni (1974) also found significant correlations between TOEFL and GPA, although the correlations were not strong (r = .270 with first-semester GPA, .336 with second-semester GPA). They decided that the TOEFL was about as useful for predicting success for nonnative English speakers as other admissions tests are for native speakers. Baldauf and Dawson (1980), after doing a prediction study of students in a teacher-training college in Papua, New Guinea, where English was the language of instruction, concluded that the MTELP was a reliable and valid predictor of general academic attainment for their students. They found correlations ranging from .33 to .74. The MTELP was also used in a validity study of 42 Cuban American students in a teacher education program at Florida International University (Freidenberg & Curry, 1981). Test scores were significantly but, in the opinion of the researchers, not strongly related (r = .41) to GPA. Ayers and Peters (1977) found a similar correlation between TOEFL and GPA (r = .40) in a study of 50 male Asian graduate students of engineering, chemistry, and mathematics. Odunzes (1982) study of 118 Nigerian students in Missouri found no significant correlation (r = .259) between the TOEFL and firstyear grades. This result led him to question the suitability of the TOEFL as a measure of English language proficiency for Nigerian students. Notwithstanding the data from the study, Odunze did not question the importance of English language proficiency to academic achievement. Ho and Spinks (1985) studied the predictive value of English language skills scores (obtained by tests of reading, writing, listening, and speaking) and a number of other measures for 230 university students in Hong Kong. Theirs was a complicated study

including many variables which were submitted to canonical correlation and multiple correlational analyses. From these analyses, it was concluded that scores on the English tests had the most predictive value, accounting for about 10% of the variance of the measures of academic success, in this case a combination of examination scores and course grades. According to Ho and Spinks, it is quite certain that students who are deficient in English (excepting those concentrating on Chinese language or nonlinguistically-dependent subjects [e.g., mathematics]) would be handicapped in their learning at the University, and might avoid choosing subjects highly dependent on English proficiency. (p. 258) To summarize, a review of the studies of the relationship between English proficiency and academic success does not reveal clear-cut answers for the ESL professional who is looking for guidance in making recommendations to admissions offices. LIMITATIONS OF ENGLISH LANGUAGE PROFICIENCY TESTS Does the seemingly erratic ability of English proficiency test scores to predict academic success mean that the tests are not valid measures of English language proficiency? Not necessarily. An impressive number of researchers and writers have concluded that the tests commonly used are valid as measures of English language proficiency, Gershman (1977) reported the findings of a Toronto Board of Education study of the validity of the TOEFL and the MTELP. The subjects were 338 Grade 13 students who reflected the diversity of ethnic, national, and linguistic backgrounds of Canadian residents: Some were native speakers who had always lived in Canada; some were nonnative speakers who had always lived in Canada; others were native English-speaking or nonnative English-speaking immigrants. The main conclusion of the study was that while the English scores did not predict GPA, they successfully identified those who had learned English as a second language. Further support for the construct validity of traditional English proficiency tests comes from the authors of a study (Duran, Canale, Penfield, Stansfield, & Liskin-Gasparro, 1985) that is critical of the TOEFL and that offers suggestions for its improvement. Despite their criticisms and suggestions, they state unequivocally that research on the test [TOEFL] has clearly demonstrated its value as an instrument for assessing the language proficiency of incoming foreign college students (p. 61).

If the TOEFL is valid as a measure of English proficiency, then the MTELP and the CELT are also very likely valid. As Palmer and Woodford (1978) have pointed out, all the high level proficiency tests [such as the TOEFL, the MTELP, and the American Language Institute at Georgetown University test] measure pretty much the same thing. As a consequence, they all tend to correlate (p. 509). Examples of this kind of correlation are many: Dizney (1965) found a correlation of .97 between the TOEFL and the MTELP in a study involving 20 students; Cervenka (1978) reported correlations in the .80s between the TOEFL and the MTELP; and J.B. Carroll (1972) reported a correlation of .91 between the TOEFL and the CELT in one study involving 29 students. In spite of this kind of evidence, some critics will argue that a different sort of English test, one that focused on testing communicative competence rather than formal knowledge, would be a better measure for college and university applicants. In a recent invitational conference called by the Educational Testing Service, a number of experts in the field gave suggestions for improving the TOEFL. Among the suggestions were (a) to provide more cohesive and coherent context for the items in the test; (b) to emphasize substance and meaning rather than form; (c) to increase the authenticity of the language of the test, making it more representative of the real-life language of the academic setting (Savignon, 1986); and (d) to devote a portion of the test for the candidates to demonstrate, in Gallegnos (cited in Larsen-Freeman, 1986) words, what they do know, not being penalized for what they dont know (p. 34). Despite these kinds of suggestions, which are based on theoretical constructs of communicative competence, the fact remains, according to Oller (1986), that everything points to the conclusion that the TOEFL is presently a fairly good measure of communicative competence (p. 149). It is possible that tests which require the participants to demonstrate that they can use English rather than simply demonstrate that they can recognize correct linguistic items would be better predictors of academic success, but it is not at all certain. Saville-Troike (1984), in a study of elementary school children, found that some students who virtually never spoke English with schoolmates did the best academically. And she found that vocabulary knowledge was the best language skill predictor of academic success. Hanania and Shikhani (1986) reported that adding an essay to the regular proficiency test used at the American University of Beirut, a test resembling the TOEFL, added only marginally to the predictive value of the scores (p. 99).

What Does the Lack of Consistently High Correlations Mean? First, it seems clear that many factors other than English language proficiency determine academic success. As Palmer and Woodford (1978) have pointed out, All TOEFL can indicate if it is working right, is, can the person function in the language? (pp. 501-502). They stress that a test designed to measure language ability is not necessarily going to predict success in an academic program. The lack of consistently high correlations between English proficiency scores and academic success may also result in part from the fact that foreign students with low levels of proficiency are not usually admitted to colleges and universities. As the TOEFL Test Manual (Educational Testing Service, 1985) explains, If an institution admits to academic work only those foreign applicants who have demonstrated a high level of language competence, one would expect that English proficiency would not be a major factor in influencing academic success (p. 16). If students with lower levels of English language proficiency were regularly admitted into academic programs, it is entirely possible that correlations between English proficiency scores and academic success would be stronger. This hypothesis receives limited support from two of the studies summarized earlier. Some of the strongest correlations between TOEFL and GPA were found by Gue and Holdaway (1973) (r = .49 and r = .59); the students in their sample had relatively low mean TOEFL scores: 424.6 and 447.8 respectively. In contrast, Light, Xu, and Mossup (1987) obtained a correlation of only .14 between TOEFL and GPA in their study, but their sample had a high mean TOEFL score: 561. Some researchers offer another explanation for the lack of consistently high correlations. Typical proficiency tests are flawed in that they do not take sociolinguistic factors into account. Odunze (1982) explained the failure of the TOEFL to predict academic performance of his Nigerian subjects by citing their strong dislike for multiple-choice tests. Similarly, Canale (1984) has claimed that performance on language proficiency tests may be influenced by individuals attitudes toward and acceptance of certain [test] tasks (p. 38). One clear conclusion that the lack of consistently high correlations between English test scores and academic success leads to is that English proficiency is only one among many factors that affect academic success. Factors Influencing Academic Success Intuitively, one would expect that general intelligence and academic skill in areas other than English language would have an

effect on academic performance. Support for this intuition is provided by Sharons (1972) finding that for the 975 foreign graduate students in his study, the best predictor was the quantitative portion of the GRE. Wilcox (1975) found that math aptitude scores, high school GPAs, and subject-matter achievement test scores all showed significant correlations with academic success, while English proficiency did not. A study conducted at the University of Maryland (Graham, 1984) showed that for the nonnative English speakers at the Baltimore County campus, math SATS and high school grade point averages were the best predictors. (Many prediction studies of American college students have found high school grades or rank in class to be better predictors than any aptitude test, often with correlations of about .50 between these measures of academic achievement and success in college [Wigdor & Garner, 1982]). Several of the researchers who have done predictive studies of English language proficiency tests have come to the conclusion that nonacademic, as well as academic, factors play a role in academic achievement. Ho and Spinks (1985) claim that nonacademic factors connected with personality and attitude may well affect academic performance and that admissions schemes perhaps ought to take these into account, as well as more traditional academic criteria. Gue and Holdaway (1973) mention a number of nonacademic factors which they believe might influence success: Motivation, homesickness, friendships, acceptance by the Western culture, and attitudes to learning, may combine to offset good language proficiency, or to overcome initially lower proficiency (p. 102). The Proper Use of Language Tests Does all this mean that English proficiency is not important to academic success in English-language colleges and universities? Not at all. What the research does show is that the relationship between English proficiency and academic success is complex and unclear and that language test scores should not therefore play a disproportionate role in admissions decisions. The Educational Testing Service (1985, p. 16) itself urges institutions not to use TOEFL scores as the sole basis for admissions decisions, not to use rigid cutoff scores, and not to use the scores for predicting academic performance. According to Palmer and Woodford (1978), it makes the Educational Testing Service very nervous when the TOEFL is used for admissions (p. 505), and they remind college administrators that it gives information about just one aspect of a

candidate. The same, of course, can be said for any language proficiency test. But while proficiency scores should not be given excessive weight, some evidence seems to show that the lower the average English proficiency is, the greater a factor it is in students success or lack of it. This may be particularly true when definitions of academic success include the amount of academic work successfully completed, rather than simply the GPA (see Light, Xu, & Mossop, 1987). In other words, it seems likely that there is a minimal level of English proficiency required before other factors assume more importance. What that minimal level is will almost certainly vary from institution to institution and, indeed, from program to program. WHAT THE ESL PROFESSIONAL CAN DO Given both the ambiguity of the relationship between English proficiency and academic success and the lack of clear guidelines as to minimal levels of proficiency required for success, what should conscientious ESL practitioners do when consulted about admissions decisions? The following suggestions are offered: 1. Establish some initial guidelines. If asked for an opinion about cutoff scores, ESL practitioners can refer to the TOEFL Test Manual (Educational Testing Service, 1985), which lists TOEFL ranges in the 900 institutions surveyed by the Educational Testing Service. (They should also share with admissions people the Educational Testing Services warning about the use of cutoff scores.) For scores of standardized tests other than TOEFL, they can consult institutions similar to their own about the score ranges used for admissions decisions. 2. Attempt to find the minimal level of English proficiency needed at the institution. To do this, ESL practitioners can monitor the academic achievement of students with limited English-speaking ability in their institution and study their scores on English proficiency measures. Palmer and Woodford (1978) have argued that each institution needs to do its own studies to establish appropriate levels; given the varying expectations and standards of different institutions, that advice seems well taken. 3. Design tests for the institution to supplement commercial tests or indeed to replace them. This last should be done only after considerable reading and with support from testing experts, however.

4. Keep in mind that visiting foreign students and limited Englishspeaking immigrants or permanent residents may be two populations with distinctly different backgrounds and characteristics. While little research has been done on these differences, some studies indicate that tests which give useful results for foreign students may be worthless for immigrants or permanent residents (Verts, 1984). 5. Continue to read relevant articles and books and to attend conference sessions dealing with English proficiency and academic success. Research on testing is an active field, and breakthroughs in testing are possible. But above all, ESL practitioners must remember that there will never be a single, ideal test and that the necessary level of English proficiency will almost certainly vary according to the particular program that a student enters and according to attributes other than English proficiency that the student brings to academic pursuits. While it may be discouraging that the whole question of English proficiency and its relationship to academic performance is as complex as it is, admissions decisions must be made. Since both common sense and empirical evidence suggest that there are varying threshold levels of English proficiency necessary for academic success in different programs, recommendations can be made with this in mind. One solution, for example, might be to make English proficiency requirements so high for all applicants that proficiency ceases to be a factor in success or lack of success. Another solution is to adapt the proficiency requirements to the number of ESL support services available on a particular campus. A third solution is to adjust the level of proficiency requirements according to the strength of an individual applicants other qualifications, such as mathematics scores or high school GPA. A fourth solution is to recommend that certain applicants be admitted provisionally, that is, with restrictions (such as a limit on the number of courses they are allowed to register for) or requirements (such as signing up for an ESL course or tutor), Undoubtedly, even with the best guidelines, errors will occasionally occur: Students will be admitted who do not succeed, and others will be kept out who could have succeeded. What conscientious ESL professionals must do is to keep up with current research, study the relationship between English proficiency scores and success on their own campuses, and make sure that their recommendations are made on the basis of careful consideration and the best information available.



THE AUTHOR Janet G. Graham (PhD, English Education, University of Maryland) is the Language Coordinator at the University of Maryland, Baltimore County, where she teaches ESL and administers ESL programs. Her primary research interests are English for specific purposes, fossilization, and writing development. She is a former president of Baltimore Area TESOL. REFERENCES Ayers, J. B., & Peters, R.M. (1977). Predictive validity of the test of English as a foreign language for Asian graduate students in engineering, chemistry, or mathematics. Educational and Psychological Measurement, 37, 461-463. Baldauf, R. B., Jr., & Dawson, R.L.T. (1980). The predictive validity of the Michigan test of English language proficiency for teacher trainees in Papua New Guinea. Educational and Psychological Measurement, 40, 1201-1205. Bostic, M.L. (1981). A correlational study of academic achievement and the test of English as a second [sic] language (TOEFL). Dissertation Abstracts International, 42, 468A. (University Microfilms No. 81-16, 851) Burgess, T. C., & Greis, N.B. (1970). English language proficiency and academic achievement among students of English as a second language at the college level. (ERIC Document Reproduction Service No. ED 074 812) Canale, M. (1984). On some theoretical frameworks for language proficiency. In C. Rivera (Ed.), Multilingual Matters, 10, 28-40. (ERIC Document Reproduction Service No. ED 240 882) Carroll, B.J. (1980). Testing communicative competence: An interim study. Oxford: Pergamon. Carroll, J.B. (1972). A comprehensive English language test for speakers of English as a second language. In O.K. Buros (Ed.), The seventh mental measurements yearbook (Vol. 1, p. 260). Highland Park, NJ: Gryphon. Cervenka, E.J. (1978). Michigan test of English language proficiency. In O.K. Buros (Ed.), The eighth mental measurements yearbook (Vol. 1, pp. 104-106). Highland Park, NJ: Gryphon. Dizney, H. (1965). Concurrent validity of the test of English as a foreign language at an American university. Educational and Psychological Measurement, 25, 1129-1131. Dulay, H., Burt, M., & Krashen, S. (1982). Language two. New York: Oxford University Press. Duran, R. P., Canale, M., Penfield, J., Stansfield, C. W., & Liskin-Gasparro, J.E. (1985). TOEFL from a communicative viewpoint on language proficiency: A working paper (Research Rep. No. 17). Princeton, NJ; Educational Testing Service. Educational Testing Service. (1985). TOEFL test manual: Princeton, NJ: Author.

Farhady, H. (1983). New directions in ESL proficiency testing. In J.W. Oller, Jr. (Ed.), Issues in language testing research (pp. 253-269). Rowley, MA: Newbury House. Freidenburg, J. E., & Curry, W.C. (1981, May). English proficiency and the bilingual university student. Paper presented at the Annual International Bilingual Bicultural Education Conference, Boston. (ERIC Document Reproduction Service No. ED 208 657) Gershman, J. (1977). Testing English as a foreign language: Michigan/ TOEFL study. Toronto: Toronto Board of Education. (ERIC Document Reproduction Service No. ED 145 693) Graham, J.G. (1984). Predictors of ESL student academic success. Unpublished manuscript, University of Maryland, Baltimore County. Gue, L. R., & Holdaway, E.A. (1973). English proficiency tests as predictors of success in graduate studies in education. Language Learning, 23, 8 9 - 1 0 3 . Hanania, E., & Shikhani, M. (1986). Interrelationships among three tests of language proficiency. TESOL Quarterly, 20, 97-109. Hatch, E., & Farhady, H. (1982). Research design and statistics for applied linguistics. Rowley, MA: Newbury House. Heil, D. K., & Aleamoni, L.M. (1974). Assessment of the proficiency in the use and understanding of English by foreign students as measured by the test of English as a foreign language (ERIC Research Rep. No. 350). (ERIC Document Reproduction Service No. ED 093 948) Ho, D. Y. F., & Spinks, J.A. (1985). Multivariate prediction of academic performance by Hong Kong University students. Contemporary Educational Psychology, 10, 249-259. Hwang, K., & Dizney, H.F. (1970). Predictive validity of the test of English as a foreign language for Chinese graduate students at an American university. Educational and Psychological Measurement, 30, 475-477. Larsen-Freeman, D. (1986). A response to S. Savignons The meaning of communicative competence in relation to the TOEFL program. In C. W. Stansfield (Ed.), Toward communicative competence testing: Proceedings of the second TOEFL invitational conference (Research Rep. No. 21) (pp. 31-37). Princeton, NJ: Educational Testing Service. Light, R.L., Xu, M., & Mossop, J. (1987). English proficiency and academic performance of international students. TESOL Quarterly, 21, 251-261. Manuel, H.T. (1962). Test of reading and number: Inter-American series. San Antonio: Guidance Testing Associates. Mestre, J.P. (1981). Predicting academic achievement among bilingual Hispanic college technical students. Educational and Psychological Measurement, 41, 1255-1264. Mulligan, A.C. (1966). Evaluating foreign credentials. College and University, 41, 307-313. Odunze, O.J. (1982). Test of English as a foreign language and first year GPA of Nigerian students. Dissertation Abstracts International, 42, 3419A-3420A. (University Microfilms No. 82-02, 657) 520 TESOL QUARTERLY

Oller, J. W., Jr. (1986). Communication theory and testing: What and how. In C. W. Stansfield (Ed.), Toward communicative competence testing: Proceedings of the second TOEFL invitational conference (Research Rep. No. 21) (pp. 104-155). Princeton, NJ: Educational Testing Service, Palmer, L. A., & Woodford, P.E. (1978). English tests: Their credibility in foreign student admissions. College and University, 53, 500-510. Savignon, S.J. (1986). The meaning of communicative competence in relation to the TOEFL program. In C. W. Stansfield (Ed.), Toward communicative competence testing: Proceedings of the second TOEFL invitational conference (Research Rep. No. 21) (pp. 17-30). Princeton, NJ: Educational Testing Service. Saville-Troike, M. (1984). What really matters in second language learning for academic achievement? TESOL Quarterly, 18, 199-219. Sharon, A.T. (1972). English proficiency, verbal aptitude, and foreign student success in American graduate school. Educational and Psychological Measurement, 32, 425-431. Shay, H.R. (1975). Effect of foreign students language proficiency on academic performance. Dissertation Abstracts International, 36, 1983A. (University Microfilms No. 75-21, 931) Slark, J., & Bateman, H. (1982). A study of non-native English speakers academic performance at Santa Ana College. ( E R I C D o c u m e n t Reproduction Service No. ED 206 368) Stover, A.D. (1982). Effects of language admission criteria on academic performance of non-native English-speaking students. Dissertation Abstracts International, 42, 4374A-4375A. (University Microfilms No. 8207, 017) Sugimoto, R.A. (1966). The relationship of selected predictive variables to foreign students achievement at the University of California, Los Angeles. Dissertation Abstracts, 28, 65A. (University Microfilms No. 678032) Verts, L.J. (1984, July). Integration of ESL/LEP students into the university. Paper presented at the Sixth TESOL Summer Meeting, Corvallis, OR. Wigdor, A. K., & Garner, W.R. (Eds.). (1982). Ability testing: Uses, consequences, and controversies (Part II). Washington, DC: National Academy. Wilcox, L.O. (1975). The prediction of academic success of undergraduate foreign students. Dissertation Abstracts International, 38, 6 0 8 4 B . (University Microfilms No. 75-12, 178)



TESOL QUARTERLY, Vol. 21, No. 3, September 1987

Teachers and Students Learning About Compliments

JANET HOLMES Victoria University of Wellington DOROTHY F. BROWN Sydney College of Advanced Education

Paying appropriate compliments and identifying them accurately is an aspect of communicative competence which may differ in a variety of ways from one culture to another. This article provides examples of misunderstandings in compliment exchanges in different cultural contexts and analyzes them as instances of pragmalinguistic and sociopragmatic failure. An analysis of pragmalinguistic and sociopragmatic norms of complimenting behavior in two English-speaking communities forms the basis of a set of exercises designed to assist learners in developing the ability to recognize and use compliments appropriately.

Considerable debate has focused on the extent to which it is possible to teach communicative competence. Some argue that communicative aspects of language, such as expressing oneself politely and participating smoothly in a conversation, are absorbed unconsciously in the process of learning more obviously teachable features (e.g., George, 1978; Jakobovits & Gordon, 1980; Rintell, 1979; Terrell, 1977). Indeed, George goes further, arguing that tolerance, not teaching, is all that is required: More people than ever before move from one language speaking area to anotherand learn the languages they need to the extent they need them. . . . It seems that there is no need for such norms to be normative, provided there is tolerance. Conversely, a sociolinguistic error is an error only when there is enough lack of tolerance to make it one. More than ever before language learning is taking place outside classrooms. (p, 9) Others suggest that these components of competence will develop naturally, provided the classroom environment is properly

structured (Munby, 1978; Vogel, 1977). Still others believe that learners should be made aware of the ways native speakers use language to understand not only the forms, but also the appropriate situations in which to use them (e.g., Edmondson, House, Kasper, & Stemmer, 1984; House & Kasper, 1981; Thomas, 1983; Wolfson, 1981). Edmondson et al., for instance, emphasize the importance of cognitive learningi.e. of knowledge about communicative norms, values and presuppositions of ones own and the target culture (p, 124), and Thomas points out that teachers should be sensitizing learners to expect cross-cultural differences in the linguistic realizations of politeness, truthfulness, etc. (p. 110). We subscribe to this latter view. This article provides data on complimenting and suggests how teachers can help ESL learners acquire the information they need to understand and use compliments in some English-speaking speech communities. Data on two speech communities are used to illustrate the sociolinguistic rules relating to complimenting behavior. The research of Wolfson (1981, 1983a, 1983b), Manes (1983), and Manes and Wolfson (1981) on complimenting behavior in U.S. middle-class adult society appears to describe norms which extend over wide areas of the United States. This is supplemented by data on compliments in the speech of white middle-class adult New Zealanders (Holmes, in press). The New Zealand data were collected by 10 students, using the ethnographic method described in Manes and Wolfson (1981). Each of the students simply noted, without selection or editing, a sequence of 20 compliments in the order in which they naturally occurred. They noted the exact words of the compliment and of the compliment response, the sex and relative status of the participants, and the context in which the compliment occurred. Though this method is not as accurate as tape-recording, it is perfectly adequate for collecting compliment sequences when phonology is not the focus of study, and it is very much more productive than attempts to record a sufficiently rich and varied sample on tape. Thus, the New Zealand corpus of 200 compliments analyzed for this study was collected from New Zealand adults in a wide range of informal contexts. Exactly 50% (i.e., 100) of the New Zealand compliments occurred between female participants, and almost a quarter more (i.e., 46) were given by females to males. The number of compliments given by males was considerably fewer. Males gave 37 (18.5%) compliments to women and 17 (8.5%) to other men. The results are presented in more detail in Holmes (1987, in press).

CROSS-CULTURAL MISCOMMUNICATION There is evidence that paying compliments is a troublesome aspect of English for learners from different cultural backgrounds.1 Wolfson (1981) describes some of the problems encountered by ESL learners in the United States. Some learners simply find it difficult to understand why, from their point of view, Americans use compliments so frequently. In Indonesian culture, for example, Wolfson reports, compliments are relatively rare and are used only among members of the educated class who have been exposed to Western customs. Malaysian students in New Zealand confirm this observation, commenting that they were very surprised and somewhat discomfited by the frequency of compliments between New Zealanders. Even within English-speaking cultures, there are differences in the frequency with which it is acceptable to give compliments, New Zealanders, for instance, tend to feel that Americans pay far too many compliments and, judging by their own norms, assume that American compliments are often insincere. In some cases, Wolfson points out, it is not even clear that different cultural groups would agree on what counts as a compliment. Comments on appearance by Americans, for example, are often interpreted by those from other cultures as criticisms, implying the addressee did not normally look good. Wolfson (1981) cites the following example: 1. Hey, whats the occasion? You look really nice today. (p. 119) A non-American addressee is likely to be hurt by the suggestion that his or her appearance is unusual and will respond accordingly. In such cases there is a risk of offense both to the addressee and to the compliment-giver, for if an intended compliment is not recognized, it will not be responded to appropriately. In New Zealand (as well as in Australia), we have noted examples of amusement, embarrassment, or offense unwittingly caused by or given to ESL learners from different cultural backgrounds. In the following example, a Malaysian male student compliments his university tutor as she enters the classroom. 2. Complimenter: You are wearing a very lovely dress. It fits you. Recipient: Ohthank you.

Responses to compliments are also an area of potential misunderstanding. Malaysian students in New Zealand comment that they find it very difficult to accept a compliment and that their preferred strategy, adopted from Malay, is to disagree. Further research is being undertaken on this component of the compliment routine (see Holmes, 1987).


This example illustrates an obvious linguistic error, a covert grammatical error, in Thomass (1983, p. 94) terms, in that the complimenter almost certainly intended to express the meaning of suits rather than of fits. However, the recipient reported being taken aback at receiving this compliment, commenting that it seemed too familiar from a young male student but that she felt it might have been acceptable from a female student. Example 3, in which a female Pakeha (New Zealander of European descent) compliments a Samoan friend whom she is visiting, illustrates another such miscommunication: 3. Complimenter: What an unusual necklace. Its beautiful. Recipient: Please take it. In this case, the complimenter was very embarrassed at being offered as a gift the object she had admired. This would have been perfectly predictable, however, to anyone familiar with Samoan cultural norms with respect to complimenting behavior. In formal settings, too, careful attention must be paid to cultural norms, as the following excerpt from a Sydney newspaper illustrates: 4. Bob Hawkes eyes grew wider and wider as he was introduced by an official of the Japan National Press Club at a news conference on Friday. The 150 or so Japanese and Australian journalists heard that the Prime Ministers golf handicap was a State secret so we should not enquire about it. His popularity among ladies and lady voters is a pronounced phenomenon and also in the past he prided himself on his large capacity for alcoholic intake. (Sydney Morning Herald, 1986, p. 1) While the Japanese official undoubtedly intended the remarks to be complimentary, Mr. Hawke was clearly disconcerted by their content in such a formal setting. The speakers lack of sociolinguistic as opposed to grammatical competence lay at the heart of Mr. Hawkes discomfiture. ANALYZING COMPLIMENTS In analyzing the causes of cross-cultural misunderstandings arising from complimenting behavior, the distinction made by Thomas (1983) between pragmalinguistic failure and sociopragmatic failure is a very useful one. Pragmalinguistic failure refers to a misunderstanding of the intended illocutionary, or pragmatic, force of an utterance, as illustrated by Example 5, in which a Tokelau womans comment to a Pakeha woman visiting her is misinterpreted by the latter as a compliment:

5. Complimenter: Youve lost a lot of weight. What have you been doing? Recipient: Thank you. Ive started jogging regularly and it seems to work. Complimenter: You shouldnt overdo it. You are looking quite thin. Although the Pakeha woman interpreted her friends comment on her weight as a compliment, it becomes clear that it was intended as an expression of concern. The basis for the misunderstanding in this case lies in different perceptions of appropriate topics for compliments. Weight loss is considered by some cultures to be a cause for concern, whereas in others it is a cause for congratulation. Even in Pakeha culture, however, severe and sudden weight loss would cause concern. Moreover, what counts as desirable and undesirable weight loss will vary from individual to individual. Though Example 5 happened to involve people from different cultural backgrounds, it might easily have occurred between native speakers with different views about the Pakeha womans ideal weight, such as her mother and herself. Accurate interpretation of illocutionary force involves knowledge of subtle cultural norms. The following two examples also illustrate pragmalinguistic failures. In Example 6, the complimented, from a culture where large families are considered an asset, is visiting a New Zealander who interprets the persons remark as critical. 6, Complimenter: What a big family you have! Recipient: Yes, but it has its advantages, too. The complimented in Example 7 is a young Chinese interpreter who is aware the addressee knows only a very few Chinese phrases. The recipient therefore believes the compliment is insincere (in her judgment, it offends against Grices [1975] maxim Be truthful), and so she assumes the interpreter is simply fishing for a compliment on her English. 7. Complimenter: Your Chinese is very good. Recipient: Oh no its not, but your English is really very good. Though such misunderstandings are possible between native speakers, they occur much more frequently and systematically between members of different cultural groups. Identifying the potential sources of predictable misunderstandings is the language teachers task. According to Thomas (1983), the language teachers concern is appropriately directed to cases in which the pragmatic force mapped by S [the speaker] onto a given utterance is systematically different from the force most frequently assigned to it

by native speakers of the target language, or when speech act strategies are inappropriately transferred from L1 to L2. (p. 99) The language teacher will need to listen for and note such cases because the best examples for a particular class will be those drawn from the personal experiences of class members as they are learning English. Sociopragmatic failure, which can also be accounted for by inadequate knowledge of relevant cultural and social values, occurs when a speaker selects an inappropriate linguistic strategy to express a speech act in a particular context. Example 2 illustrates this kind of miscommunication. The utterance uses a perfectly acceptable linguistic strategy for expressing a compliment; the recipient had no trouble in interpreting its intended pragmatic force. The problem was that the compliment was perceived as too intimate, given the social context and the relationship between the participants. Thomas (1983) describes causes of sociopragmatic failure in terms of Brown and Levinsons (1978) universal criteria for assessing the relative politeness of different speech acts: It is cross-cultural mismatches in the assessment of social distance, of what constitutes an imposition, of when an attempt at a facethreatening act should be abandoned, and in evaluating relative power, rights, and obligations, etc., which cause socio-pragmatic failure. (p. 104) Knowing whether a compliment is appropriate at all, as well as which linguistic strategy to select to express it, is part of the communicative competence learners need to acquire. This is illustrated by Example 8, in which a 4-year-old girl on a bus loudly and clearly addresses an elderly woman whom she does not know. 8. Complimenter: Thats a nice hat. Recipient: Ohthank you. The recipient was visibly embarrassed, as was the 4-year-olds mother. The child selected a perfectly appropriate linguistic strategy for encoding a compliment it was her assessment of the relevant social factors in the situation which betrayed her communicative inexperience. She had not yet learned to take proper account of factors such as the relationship between the participants and the publicness of the situation. Moreover, compliments on some topics may be more acceptable than others in a particular context. For example, a male learner who expresses appreciation to a teacher after a lesson he has enjoyed is unlikely to cause embarrassment of the sort caused by a comment 528 TESOL QUARTERLY

on the teachers appearance, as in Example 2. Wolfson (1983a) suggests that in the United States, the safest compliments to offer to strangers relate to possessions (e.g., Thats a beautiful car) or to some aspect of performance intended to be publicly observed (I really enjoyed your talk yesterday) (p. 90). Judging which possessions and which aspects of performance are appropriate topics for compliments, however, may require extensive knowledge of the relevant culture. Distinctions in Maori culture between possessions which are tapu (i.e., sacred, invested with a power that can be positive or negative) and those which are noa (i.e., not tapu), for instance, affect the kind of comments which may be appropriately made about them, and a thorough understanding of the distinction and its significance requires a deep knowledge of Maori culture (Benton, 1985; Metge, 1967, 1979). Pragmalinguistic Competence Manes and Wolfson (1981) found that compliments are remarkably formulaic speech acts: A very small number of lexical items and syntactic patterns account for the great majority of the 686 compliments in their corpus. This finding was replicated in the New Zealand corpus of 200 compliments analyzed for this study. Table 1 summarizes the three most commonly occurring syntactic patterns in the U.S. and New Zealand data. These three patterns account for 85% of the U.S. and 78% of the New Zealand data.
TABLE 1 Relative Frequency in Percentage of Three Syntactic Patterns Used in Compliments



Compliments also draw on a very restricted range of lexical items. Wolfson (1984) notes that 80% of the compliments in her corpus depended upon an adjective to carry the positive semantic load (p. 236). Moreover, while the number of positive adjectives which may be used is virtually unlimited . . , two thirds of all adjectival compliments in the corpus made use of only five adjectives: nice, good, beautiful, pretty and great (p. 236). In the New Zealand data, the five most frequently occurring adjectives were nice, good, beautiful, lovely, and wonderful. Most of the nonadjectival compliments also depended upon just a few semantically positive verbs (like, love, enjoy, admire, and b e impressed by), with like and love alone accounting for 86% of the American data and 80% of the New Zealand data. The similarities between the U.S. and New Zealand data confirm Wolfsons claim that the formulaic nature of this speech act may extend to other English-speaking communities. Another noteworthy aspect of pragmalinguistic competence involves the topics which may appropriately serve as the focus of a compliment. To be interpreted as a compliment, an utterance must refer to something which is positively valued by the participants and attributed to the addressee. Once again, the U.S. and New Zealand data suggest that though the range of possible topics is very broad, the vast majority of compliments refer to just a few general topics. For the U.S. data, Wolfson (1983a) says, with respect to topic, compliments fall into two major categoriesthose having to do with appearance and those which comment on ability (p. 90). Manes (1983), discussing the first of these categories, comments on the overwhelming number of compliments on personal appearance, most particularly clothes and hair-dos (p. 98) and points out that aspects of appearance which are the result of deliberate effort (p. 99) are particularly likely to attract comment, as are new items. The second most frequent topic in the U.S. data, identified by Wolfson as ability, is described more fully by Manes as the quality of something produced through the addressees skill or effort: a well-done job, a skillfully-played game, a good meal (p. 101). These two topics are also the most commonly occurring ones in the New Zealand data: Compliments on appearance account for 45.0% of the total, and compliments on ability or performance make up 27.5% of the data. Compliments on possessions (e. g., I love your new bike) and those having to do with personality/friendship (e.g., That was kind) account for 10.5% and 13.5%, respectively. Compliments tend to occur at the openings and closings of speech events, often preceded by greetings and followed by

farewells, as illustrated in Example 9, in which two colleagues encounter each other for the first time one workday 9. Recipient: Hi, hows things? Complimenter: Fine. Youre looking very smart today. Recipient: Thanks. I decided it was time to splash out on something new. Compliments may also occur at transition points within a speech event, and there are a number of other interesting restrictions on where compliments may occur (see Holmes, 1987). While research is needed on this aspect of pragmatic knowledge, it appears that a compliment in the middle of a conversation requires special justification. This is illustrated in Example 10, in which the complimented interrupts two students who had been discussing their work outside the library. 10. Complimenter: . . . its really time I finished it thoughoh I just realized youve had your hair cut. Recipient: Yes, I had it done yesterday. Complimenter: I knew there was something different but I couldnt quite work out what. It looks great. Recipient: Well, it feels a lot better I must say. Finally, it should be noted that compliments can serve as bivalent or plurivalent speech acts, expressing more than one illocutionary or pragmatic force (Thomas, 1985). In general, they serve to reinforce desired behavior (Manes, 1983). Hence, compliments are often used in English-speaking communities to express encouragement. Youre doing well, good work, and well done are typical of compliments expressed in classrooms. Example 11, in which a primary school teacher addresses a class of 5-year-olds, illustrates a particularly common device used by teachers, namely, a compliment that provides encouraging feedback to the recipient while also serving as an indirect directive to the rest of the class: 11. Complimenter: Marys sitting up nicely. Recipient: [Nonverbal response: Mary sits up even straighter, looking pleased.] Compliments can also be used as expressions of gratitude, as illustrated in Example 12, in which a guest compliments the host when leaving after a dinner party. 12. Complimenter: I loved the tacos. Ill be back tomorrow for some more. Recipient: Hope they werent too hot.

Wolfson (1983a) points out that compliments serve as social lubricants and often accompany or even replace other speech act formulas, such as apologies, thanks, and greetings. They may also precede, and so soften, a criticism. Wolfson suggests that the exact conditions under which a compliment may replace other speech functions, such as expressions of gratitude, are by no means selfevident, and she notes that social factors such as role and status relations must be carefully considered. Sociopragmatic Competence The selection of appropriate topics and the use of compliments to perform additional functions depend on factors which are relevant to developing sociopragmatic competence, too. As Thomas (1983) has remarked, misunderstandings are not always clearly attributable to pragmalinguistic, as opposed to sociopragmatic, failure: There is some overlap between the concepts. Cultural beliefs and social values become the major focus, however, when we consider how native speakers select appropriate compliments in particular social contexts. Who compliments whom in which contexts and on what topics? When responding to this question, we must treat our data circumspectly, since compliments rnay be easier to collect in some contexts than in others and there is no guarantee that the corpora are representative in this respect. We have little data on childrens interactions, for example, and most of the data available have been collected by female researchers. Nevertheless, some generalizations seem to be justified. Wolfson (1983a) points out that the overwhelming majority of all compliments are given to people of the same age and status as the speaker (p. 91). The New Zealand data support this observation. As mentioned above, compliments tend to occur at certain points in interaction, and when colleagues or friends meet for the first time on a particular day, an exchange of compliments relating to appearance as part of the greeting exchangee.g., I like your dress; what a wonderful jersey; you look niceis likely. Compliments also occur in encounters between status unequals, however. Contrary to the popular view that the participant with lower status will be more likely to use compliments to flatter or manipulate, Wolfson (1983a) points out that the great majority of compliments which occur in interactions between status unequals are given by the person in the high position (p. 91). Similarly, in the New Zealand data it was interesting to find that compliments upwards were fewer than those downwards. Compliments

upwards tended to occur only when the participants knew each other reasonably well and the complimented was often a mature rather than a young person. Compliments to someone of superior status seem to require some confidence on the part of the complimenterpresumably to counteract the possibility of a negative interpretation. Wolfson also notes that compliments downwards were often work related and were twice as likely to focus on ability or performance as on appearance or possessions, whereas the reverse was the case between status equals or for compliments upwards. This tendency was observable in the New Zealand data too and partly explains why teachers report feeling uncomfortable or patronized by young overseas students who compliment them on their performance regularly and, it seems to the teacher, profusely. As Example 2 suggested, the sex of participants is an important social factor in complimenting behavior. Part of the sociopragmatic knowledge that native speakers use in interpreting and expressing compliments appropriately involves an awareness of the relevance of the sex of the participants. Wolfson (1983a) comments that in the American data women appear both to give and receive compliments much more frequently than men do (p. 92). This was certainly true for the New Zealand data as well. Women gave 73.0% of all the compliments recorded (50.0% to other women and 23.0% to men) and received 68.5% of them (50.0% from other women and 18.5% from men). Compliments between males were relatively rare (only 8.5%). While this may have been partly a reflection of the fact that the researchers were female and thus may not have been present in contexts where male-male compliments occurred frequently, its general accuracy is confirmed by Wolfsons data and by discussions with male students and colleagues. Another pattern which emerges clearly from both the U.S. and the New Zealand data is the tendency for females to receive compliments relating to their appearance. Of all the compliments women received in the New Zealand data, 47% relate to aspects of their appearance (see Table 2). New Zealand men also receive compliments on their appearance (40% of all compliments they receive), but it is interesting to note that the vast majority of these (88%) are given by women. The U.S. pattern appears to be different. Though men rarely compliment each other on appearance in either community, the appearance of American men seems not to be an appropriate topic of compliments from men or from women. Wolfson (1983a) comments that only when the male is much younger than the female does this occur at all, and in general she says, there seems to be a rather strong if not

TABLE 2 Compliment Data Analyzed According to Topic and Sex of Participants Personality/ ComplimenterAbility/perrecipient Appearance formance Possessions friendship
Female-female Female-male Male-female Male-male Total 55 22 10 3 90 24 9 20 2 55 11 5 5 21 10 10 2 5 27

5 2 7

100 46 37 17 200

categorical constraint against the giving of appearance-related compliments to higher-status males, especially in work-related settings (p. 93). Clearly, there are decided patterns concerning acceptable topics for compliments involving both the relative status and sex of the addressee and the complimented. Complimenting ones male boss on his appearance in America is likely to cause embarrassment or elicit disapproval, and complimenting a male on his clothes or hairstyle is likely to attract attention in either community if the speaker is male. CLASSROOM EXERCISES ON COMPLIMENTS The data analyzed above provided the basis for the following sets of exercises, which can help learners to develop skill in recognizing compliments, in identifying the appropriate topics and contexts for compliments, and in interpreting their functions appropriately. The first group of exercises outlined below focuses on pragmalinguistic aspects of complimenting, and the second group aims to develop sociopragmatic skills, though the distinction is of course by no means absolute. In general, the earlier exercises give the teacher more control, whereas the later ones encourage students to take the initiative in developing their sociolinguistic awareness. Some may feel that the exercises place a great deal of emphasis on talk about language at the expense of practice in using language, a criticism that was directed at the grammar-translation method and one that led to the popularity of more direct methods of language teaching. In teaching students how to use language appropriately, however, there is very good reason to encourage them to reflect on the reasons for selecting one utterance rather than another. It is

important, as Thomas (1983) says, that teachers should develop a students metapragmatic abilitythe ability to analyse language in a conscious manner (p. 98). Unless learners pay conscious attention to the relevant social factors in a particular context, they are likely to lapse automatically into the norms of their native language and culture and may thereby cause unintended offense. The kind of exercises we provide are of precisely the kind recommended by Candlin (1976, p. 251), House and Kasper (1981, p. 184), and Thomas (1983, p. 98). They are aimed at achieving sociopragmatic consciousness-raising rather than simply providing unreflecting practice of particular linguistic formulas. Developing Pragmalinguistic Competence The formulaic nature of complimentstheir syntactic and lexical predictabilitymakes them attractive ESL teaching material and provides an easy solution to the problem of how to express this speech act in English. Though indirect compliments may take a limitless variety of forms, they are proportionately so infrequent (less than 5%) that learners can legitimately use a formulaic linguistic strategy without this suggesting inadequate pragmatic competence. Hence, this aspect of pragmalinguistic competence can be taught in the same way as any other linguistic formulas, such as greetings, leave-takings, and expressions of gratitude and apology. Since just three formulas will provide for learners basic needs in this respect, the learners efforts can largely be directed at the more complicated aspects of using compliments. Exercise 1: Learning compliment formulas Directions: Distribute a copy of the chart in Table 1 and the exercise below to each student. Instruct students to identify, by writing the formula number in the space provided on the exercise sheet, which formula, if any, is being used and also to indicate the relative frequency of that formula in the New Zealand data. Relative Formula frequency number 1. You look very nice. 2. Thats a good essay. 3. I really like your hair. 4. He seems pretty unreliable. 5. I really love your garden. 6. You were so kind. 7. Thats a very nice cake you made. 8. I like your view. 9. Your new house is looking great. 10. That skirt is splendid.

Another aspect of handling compliments well which involves linguistic rather than social knowledge is coping with the collocational complexities of intensifiers. Second language learners often have problems with the collocations of intensifiers and adjectives that are so crucial to compliment formulas. Some collocations, such as very excellent and very perfect, are unacceptable because in general very should not occur with nongradable or nonsalable adjectives. Other intensifiers are slippery, in that they vary in meaning with scalable and nonsalable adjectives (e.g., quite perfect versus quite good), as well as according to the relative positions of adjectives on a scale (e.g., quite beautiful versus quite pretty). In each of the preceding pairs of examples, quite in the first member of the pair would generally be interpreted as intensifying the meaning, whereas in the second it attenuates the meaning of the following adjective, A Samoan bridegroom illustrated the confusion learners can experience with such words when he thanked a prestigious New Zealander for his speech by saying Thank you for your quite good speech. Though quite can substitute for very or really before words like outstanding or wonderful, it cannot be used as an intensifier before adjectives at a less extreme point on the scale, such as good. In fact quite is probably best avoided by learners initially. To raise students awareness of such collocational problems, the following exercise is useful. Exercise 2: Coping with compliment collocations Directions: Group students in pairs, and give each pair a copy of the chart in Figure 1. Instruct students to indicate with a check mark which intensifiers can precede which adjectives. Ask them if they need to know what is being referred to before they can make these judgments. Upon completion, each pair of students should check their answers with another pair, and then discuss their results with the class as a whole.
FIGURE 1 Collocations of Intensifiers and Adjectives nice good beautiful pretty great lovely wonderful kind

very really just absolutely pretty so


Distinguishing between compliments and other speech acts which may use similar linguistic formulas is an important pragmalinguistic skill the learner must develop. Manes and Wolfson (1981) comment, for example, that compliment formulas can be adapted with minimal effort to a wide variety of situations in which a favorable comment is required or desired (p. 123). Although it is true that simply substituting an appropriate noun phrase is often all that is needed, it is also necessary for the learner to know which noun phrases are appropriate topics of compliments. The following exercise provides examples which will generate discussion of the crucial clues to whether or not an utterance is intended as a compliment. Relevant factors include the topic of the utterance, its placing in the discourse, and its relevance to the addressee, as discussed earlier. Exercise 3: Identifying possible compliments Directions: Ask students to consider the following utterances and whether they could serve as compliments. Tell them to note the clues they use in making their decision and to provide a context to justify it. Before students begin the exercise, present the following example: Thats a neat bike. Explain that this could be a compliment, since it fits Formula 3. To function as a compliment, it would have to be addressed to the owner of the bike or to someone who could be considered responsible for it in some way. It would thus refer to a common topic of complimentspossessions. It could be said by one friend to another on the first occasion that the complimented saw the bike. 1. I simply adore pavlovas. 2. Thats a very nice thing to say. 3. You look cheerful. 4. We should do this more often. 5. Youre very rich. 6. I love fish and chips. 7. Thats a good piece of work. 8. I really enjoy opera. 9. That car looks terrifically expensive. 10. Your parents are extremely old. 11. This picnic was a very good idea. 12. Youre really well organized. Each of the preceding statements could of course be qualified, and it is precisely this which generates valuable discussion in the classroom. Exercise 4 is a similar exercise which explores the plurivalent nature of speech acts. Exercise 4: Considering plurivalent speech acts Directions: Ask students to consider how many different speech acts each of the following utterances could serve. Instruct them to

provide relevant contextual information to support their different interpretations. (The utterances in Exercise 3 can also be used for this exercise. ) The example of What neat writing youve done can be offered, with compliment from teacher to child given as the context. Point out that this utterance also acts as praise serving as encouragement to maintain good work and serves as an indirect directive to other pupils to reach similar standards. 1. I really admire your energy. 2. Thats a very bright skirt. 3. You look very happy this morning. 4. David's really good at washing dishes. 5. Your house is nice and big for entertaining. 6. How lovely to see you. 7. Thats a very clever idea. 8. Your children are so well behaved. 9. That was a beautiful meal. 10. Youre nice to have around. Developing Sociopragmatic Competence Developing awareness of acceptable topics of English compliments will lead to some discussion of the differences between the cultural and social values of the learner and the English-speaking community in which the learner wants to communicate. As a starting point, the following exercise is a useful way of raising awareness of the most frequent topics used as a focus for compliments in Enghsh-speaking cultures as well as providing an opportunity for social interaction. In addition, it exposes students to genuine sociolinguistic data in a commonly encountered format and thus provides them with practice in interpreting authentic material, a skill of great value to many adult learners. Exercise 5: Noting common compliment topics Directions: Group students in pairs, and give one member of each pair a copy of the table in Figure 2 and the other a copy of the table in Figure 3. Each table has five missing figures, which are replaced by a question mark. Instruct students to ask each other questions to complete their table but not to show their table to their partner. Present the following sample questions to students: 1. How many compliments were analyzed in total in this study? 2. How many of the compliments were about appearance? 3. Which topic is the most common one for paying compliments about? When the students in a pair have completed both tables, they should check them against the teachers master copy and then together answer these questions:

FIGURE 2 Compliment Data Analyzed According to Topic and Sex ComplimenterAbility/perPersonality/ recipient Appearance formance Possessions friendship
Female-female Female-male Male-female Male-male Total 55 22 10 3 90 24 ? 20 2 55 11 5 5 ? 10 ? 2 5 27

5 2 7

100 46 37 ? ?

FIGURE 3 Compliment Data Analyzed According to Topic and Sex ComplimenterAbility/perPersonality/ recipient Appearance formance Possessions friendship



4. What do males compliment each other about most frequently? Give one example of such compliments. 5. What percentage of all compliments are given by females to other females? 6. Why do you think females compliment males more often than males compliment females in the New Zealand data? Would this pattern be the same in your society? 7a. Collect 10 examples of compliments from short stories in a popular magazine such as the Womens Weekly, and classify them using the categories provided in the table. b. Do your data show the same distribution by topic and sex of participants as the New Zealand data? If not, why do you think your data are different? c. Do you think similar compliments would be used in your own cultural community? Information concerning the frequency of compliments between different participants and in different contexts can be very useful to the ESL learner, since it reflects the values and cultural assumptions of the community. The ESL teacher can use such information to

develop learners awareness of similarities and differences in acceptable complimenting behavior in their own language compared with English. Learning to be a good participant-observer is an invaluable skill for the language learner to develop, as Saville-Troikes (1982) comment about children suggests: Children are essentially participant-observers of communication, like small ethnographers, learning and inductively developing the rules of their speech community through processes of observation and interaction (p. 205). In addition, adult learners bring their superior cognitive and reflective skills to such observation. The following exercise suggests some ways in which teachers can make use of these strengths and assist students to become aware of cultural differences in complimenting. Exercise 6: Collecting compliment data Directions: The aim of this exercise is to help students learn how to collect their own data so that they can become aware for themselves of the contexts in which compliments occur and the appropriate topics and possible functions of compliments. The teacher should provide good examples initially in order to give students practice in noting the relevant features. The following data sources are useful and easily gradable in terms of their difficulty for students. 1. Transcripts of spoken interaction from inside the classroom. These may include praise which illustrates compliments being used for encouragement by the teacher (Thats good work), compliments linked to greetings between students (Hi. Youve had your hair done. It looks great), or compliments linked to farewells (Youve worked well today. See you tomorrow). 2. Selected written material such as magazine stories and novels. 3. Selected video sections of television situation-comedies and advertisements for analysis and discussion in class. 4. Unmonitored television or film material to be reported back to class. 5. Naturally occurring data in contexts outside the classroom to be reported back to C1aSS. Students collecting naturally occurring data should be instructed to make a record of the next 20 compliments they hear, using the format in Figure 4 to note relevant information. This should include the exact linguistic form of the compliment as well as any other contextual information they consider relevant to interpreting it. In some teaching contexts it may be possible, once students have practiced collecting English compliment data, to use the format in Figure 4 to collect compliments in the students mother tongues. Where the local language is not English or where a large group of learners use the same mother tongue, for example, such data can be

FIGURE 4 Recording Compliment Data

Comptirnent feature Linguistic form Response Topic Context Complimenter age/sex/role Recipient age/ex/role Discourse position Alternative function

Example What a great hat. Thanks I like it too. Appearance: clothes Early winter morning in the street Male student, age 24, friend of R. Male student, age 20, friend of C. Following a greeting Substitute for greeting

Note The categories in this table can obviously be modified or extended to suit learners and teachers needs.

collected and used for comparative purposes. This approach is extremely fruitful in focusing on and generating discussion of sociocultural variables which differ in emphasis between cultures. A further productive source of comparative data is role playing in the classroom, which can also provide information about possible areas of sociolinguistic interference. Moreover, role plays provide the teacher with a great deal of control over the complexity of the material. Gubbay (1980) provides examples of how role play simulates real life situations in the classroom under controlled conditions (p. 3). The teacher can first give compliments, thus controlling the range used and providing students with practice in respondinga less demanding role. Then students can practice giving compliments to each other. The kind of role play which is appropriate will depend on the composition of the class. Role play should draw on the out-of-class experience of students. Exercise 7 provides an example of how role play may be used to encourage adult students in a second language situation to reflect on the number of variables that are relevant to selecting an appropriate compliment and lead them toward an awareness of the ways these variables may interact. The exercise would need to be adapted for use with different students.

Exercise 7: Developing knowledge of factors relevant in role play Directions: Students should be given a list of situations, such as the three provided below. They should make notes on the range of possible topics and linguistic formulas for compliments in each situation, and then consider how, leaving the setting the same, these would change if variables such as sex, age, relative status of participants, and degree of familiarity are changed. Situation lMale teacher to 13-year-old female pupil in classroom Would there be any alteration in the possible topics and kinds of compliments which would be appropriate if (a) both participants were male, (b) both participants were female, (c) both participants were adult, or (d) the two met each other in the supermarket? Situation 230-year-old woman to male acquaintance she has met in shopping center Would there be any alteration in the possible topics and kinds of compliments which would be appropriate if (a) both participants were male, (b) both participants were female, (c) addressee was complimenters boss, or (d) addressee was 6 years old? Situation 3Elderly male shop assistant to unfamiliar middle-age female customer Would there be any alteration in the possible topics and kinds of compliments which would be appropriate if (a) the participants were both in their teens, (b) the customer was the local mayor, or (c) the participants were very good friends? Since native speakers appear to have reliable intuitions about the most commonly occurring formulas which may be used in compliments, role plays can be used to elicit two different kinds of data from ESL learners: data in English and data in the learners native language. These data can form the basis for valuable discussion, Using English can provide information on the extent to which the learners have acquired the formulas being discussed, and role plays using the learners mother tongue can provide information on the syntactic and lexical constraints which characterize compliments in that language. This information can then be used for comparative purposes both in relation to pragmalinguistic and sociopragmatic norms. Situations in which learners feel compliments would be inappropriate in their native culture, for instance, are likely to cause comment which can lead to useful insights for both teachers and learners. In selecting role plays, it is obviously important, as this discussion has suggested, to ensure that a range of role and status relationships are included, as well as opportunities for same-sex and cross-sex interactions. Once learners have collected data from a variety of sources, they should feel confident enough to try out what they have learned. At

this stage students can practice giving compliments and be secure in the knowledge that even if they make some mistakes, their positive intentions will generally be clear to their recipients and should generate the goodwill so valuable to learners in an unfamiliar culture. Practicing giving compliments should be a pleasant experience for all concerned. CONCLUSION Teaching sociolinguistic competence is by no means a straightforward task. As Littlewood (1983) points out, a communicative approach is concerned with the internal needs of the learner to relate to the language, integrate it with his own cognitive make-up, use it to express his own self, and so on (p. 94). Learners must also be able to integrate what they are learning with their own cultural and social values. By being too prescriptive in terms of what constitutes appropriate linguistic behavior, teachers may deprive learners of the opportunity to find ways of using the second language which they personally find comfortable. Teaching about sociolinguistic competence requires sensitivity, since sociopragmatic decisions are social before they are linguistic, and while foreign learners are fairly amenable to corrections which they regard as linguistic, they are justifiably sensitive about having their social . . . judgment called into question (Thomas, 1983, p. 104). On the other hand, the ESL teacher must provide information to learners so that they may choose how to express themselves and do so without unintentionally giving offense. Thomas (1983) expresses this point well: It is the teachers job to equip the student to express her/himself in exactly the way s/he chooses to do sorudely, tactfully, or in an elaborate polite manner. What we want to prevent is her/his being unintentionally rude or subservient. (p. 96) The approach we have taken in this article aims to provide information on norms and patterns that have been observed, as well as suggestions on how learners can make use of this information to extend their sociolinguistic competence. Learners can then decide whether to break the rules and so deliberately cause offense or amusement. They can also interpret the significance of compliments accurately. A laissez-faire, or osmotic, approach, in which the teacher expects students to simply pick up or absorb relevant knowledge without explicit teaching, risks disempowering learners, depriving them of choice and sophistication in their use of English.

Finally, it is important to stress that much more research is needed to assist ESL teachers and learners to develop competence in using language appropriately. While it is clear, for instance, that some aspects of complimenting behavior differ greatly from one community to another, the research available paints with a very broad brush. Finer analysis is needed. In many societies, for example, it is acceptable to compliment people on their possessions, but what counts as a possession varies from one culture to another. In New Zealand, for instance, compliments to a male on his wife which reflect a view of her as a possession would be considered unacceptable, except perhaps in jest between good friends; even then, not all women would be willing to see the joke! Comments on appearance are also subject to further refinement. It is not acceptable to compliment an acquaintance on simply any aspect of appearance. I like your new false teeth, for example, would be appropriate only between family members or very close friends. Comments on age or wealth are acceptable in some cultures, but not in others (Gao, 1984). Further research along these and other lines, adapted and integrated into exercises of the kind outlined in this article, will improve the quality of the information we can offer ESL learners.

ACKNOWLEDGMENTS We would like to thank the following students for their contributions to the data: S. Cotterall, M. Cranko, M. Davey, H. Drummond, G. Free, S. Gibbons, C. McCausland, F. McMichael, L. Minshall, and L. Sheung. THE AUTHORS Janet Holmes is a Reader in Linguistics at Victoria University of Wellington. She has published a number of articles which combine her interests in sociolinguistics and applied linguistics. As well as research on compliments, she has also researched sex differences in language use. Dorothy F. Brown trains teachers of English as a second language at the Sydney College of Advanced Education. She has published in the area of applied linguistics and is interested in the relevance of sociolinguistics for migrant education in Australia.

Benton, R.A. (1985). Maori, English and Maori English. In J.B. Pride (Ed.), Cross-cultural encounters: Communication and mis-communication (pp. 110-120). Melbourne River Seine. Brown, P., & Levinson, S. (1978). Universals in language usage: Politeness phenomena. In E.N. Goody (Ed.), Questions and politeness: Strategies in social interaction (pp. 56-289). Cambridge: Cambridge University Press. Candlin, C.N. (1976). Communicative language teaching and the debt to pragmatics. In C. Rameh (Ed.), Georgetown Universitg Round Table on Languages and Linguistics 1976 (pp. 237-256). Washington, DC: Georgetown University Press. Edmondson, W., House, J., Kasper, G., & Stemmer, B. (1984). Learning the pragmatics of discourse: A project report. Applied Linguistics, 5, 113-127. Gao, W. (1984). Compliment and its reaction in Chinese and English cultures. Working Papers in Discourse in English and Chinese (pp. 3237). Canberra: Canberra College of Advanced Education. George, H.V. (1978). Asian ELT: The relevance and irrelevance of sociolinguistics. Regional English Language Centre Journal, 9 (l), 1-12. Grice, H.P. (1975). Logic and conversation. In P. Cole & J.L. Morgan (Eds.), Syntax and semantics 3: Speech acts (pp. 41-58). New York: Academic Press. Gubbay, D. (1980). Role-play: The theory and practice of a method for increasing language awareness. Southall, England: National Centre for Industrial Language Training. Holmes, J. (1987). Compliments and compliment responses in New Zealand English. Manuscript submitted for publication. Holmes, J. (in press). Sex differences and language use in the ESL classroom. In B.K. Das (Ed.), Communication and learning in t h e classroom community. Singapore: Regional English Language Centre. House, J., & Kasper, G. (1981). Politeness markers in English and German. In F. Coulmas (Ed.), Conversational routine (pp. 157-185). Mouton: The Hague. Jakobovits, L., & Gordon, B. (1980). Language teaching vs. the teaching of talk. International Journal of Psycholinguistics, 6 (4), 5-22. Littlewood, W.T. (1983). Contrastive pragmatics and the foreign language learners personality. In C. Brumfit (Ed.), Learning and teaching languages for communication (pp. 90-97). London: Centre for Information on Language Teaching. Manes, J. (1983). Compliments: A mirror of cultural values. In N. Wolfson &E. Judd (Eds.), Sociolinguistics and language acquisition (pp. 96-102). Rowley, MA: Newbury House. Manes, J., & Wolfson, N. (1981). The compliment formula. In F. Coulmas (Ed.), Conversational routine (pp. 115-132). Mouton: The Hague.

Metge, J. (1967). The Maoris of New Zealand: Rautahi. L o n d o n : Routledge and Kegan Paul. Metge, J. (1979, January). The translation of culture. Paper presented at the Symposium on Cross-Cultural Aspects of Cognition, 49th Australian and New Zealand Association for the Advancement of Science Congress, Auckland. Munby, J. (1978). Communicative syllabus design. C a m b r i d g e : Cambridge University Press. Rintell, E. (1979). Getting your speech act together: The pragmatic ability of second language learners. Working Papers on Bilingualism, 17, 97-106. Saville-Troike, M. (1982). The ethnography of communication. Oxford: Basil Blackwell. Sydney Morning Herald staff. (1986, May 19). Column 8 [recurring feature]. Sydney Morning Hearld, p. 1. Terrell, T.D. (1977). A natural approach to second language acquisition and learning. Modern Language Journal, 61, 325-337. Thomas, J. (1983). Cross-cultural pragmatic failure. Applied Linguistics, 4, 91-112. Thomas, J. (1985). Complex illocutionary acts and the analysis of discourse (Lancaster Papers in Linguistics No. 11). Lancaster, England: Lancaster University. Vogel, K. (1977). Communicative competence as a performance objective and its realization in foreign language teaching. System, 5, 46-57. Wolfson, N. (1981). Compliments in cross-cultural perspective. TESOL Quarterly, 15, 117-124. Wolfson, N. (1983a). An empirically based analysis of complimenting in American English. In N. Wolfson & E. Judd (Eds.), Sociolingui.sties and language acquisition (pp. 82-95). Rowley, MA: Newbury House. Wolfson, N. (1983b). Rules of speaking. In J.C. Richards (Ed.), Language and communication (pp. 61-87). London Longman. Wolfson, N. (1984). Pretty is as pretty does. Applied Linguistics, 5, 236-244.



The TESOL Quarterly welcomes evaluative reviews of publications of relevance to TESOL professionals. In addition to textbooks and reference materials, these include computer and video software, testing instruments, and other forms of nonprint materials. Edited by VIVIAN ZAMEL
University of Massachusetts/Boston

Bilingual and ESL Classrooms: Teaching in Multicultural Contexts Carlos J. Ovando and Virginia P. Collier. New York: McGraw-Hill, 1985. Pp. xi+ 354. Bilingualism Through Schooling: Cross-Cultural Education for Minority and Majority Students Arnulfo G. Ramrez. Albany: State University of New York Press, 1985. Pp. xiii + 275. Bilingual education has long been in need of a comprehensive introductory text that succeeds at sorting out the multitude of complex issues inherent in the field. The years since the publication of two early volumes on bilingual education, Anderson and Boyers Bilingual Schooling in the United States (1970) and Cordascos Bilingual Schooling in the U.S. (1972), have been characterized by profound changes in every area related to the instruction of limited English proficient students in American schools. During this period, those who teach courses in bilingual education have relied on groups of readings on such topics as sociohistorical perspectives, law and legislation, language acquisition, language teaching and assessment, the influences of culture on learning, and the implementation of instructional programs. To that end, we have developed our own collections of readings for course use and utilized edited volumes of such readings (Center for Applied Linguistics, 1977; LaFontaine, Persky, & Golubchick, 1978; Trueba & Barnett-Mizrahi, 1979) to present the issues. There has clearly been a need for a unified text. Thus, it is encouraging that there have been two recent attempts to develop primary texts on bilingual education.

An introductory course on the nature of bilingual schooling in the United States must include a large amount of information from a number of differing fields. This might certainly help to explain the dearth of texts which can adequately address the needs of teachers preparing to work with language minority populations. The ideal text must cover substantial amounts of information on the bilingual learner, issues in L1 and L2 instruction and proficiency, sociocultural influences on school achievement, school/community variables, and instructional programs and strategies. In addition, the treatment of the historical and international development of bilingualism and bilingual schooling must set the stage for discussion of social and political implications of bilingualism and bilingual education in the United States. Clearly, the major task is to organize such material in a manner that informs the reader, ties the information together, and leads the reader to other sources for further investigation. How well do these volumes accomplish such a comprehensive task? The Ramrez volume falls short of this goal, while the Ovando and Collier text may in fact be one of the first texts in recent years to offer comprehensive and current exposure to the issues of bilingual education. Ovando and Collier indicate in the preface that their volume is directed to bilingual and ESL teachers, that the text combines theory and research with practical classroom applications (p. ix), and that as an authored text, it seeks to avoid the difficulties inherent in anthologies. Finally, they emphasize that language and culture, integral components in bilingual and ESL classrooms, form the foundation of their discussion of the options for instruction offered both language minority and language majority students. By contrast, the preface of the Ramez text makes no mention of the intended audience. Listing the variety of topics presented in the book, Ramrez states that the many studies reviewed vary . . . but each contributes to our understanding of the many dimensions of language and bilingual schooling (p. xii). The book jacket describes it as a comprehensive survey of bilingual education. An examination of each of these books must address three primary questions: (1) How does the volume present and organize the myriad topics that it must include? (2) Does the work illuminate the issues surrounding bilingual education in the United States and assist the reader in beginning to see the complex interrelationships between social, political, economic, and linguistic variables that impinge on the education of language minority students? and (3) Because the topics cannot be covered in depth,

does the book cover the basic issues and also lead the reader to original sources, other readings, and primary documents? Bilingualism Through Schooling contains a large amount of information. Material has been gathered from research studies, government reports, doctoral dissertations, and conference proceedings, with the result that skeletal reviews of each topic are presented, organized by chapter subheadings that are often not matched by their content. Such individual subheadings (the 214page book, excluding appendixes, contains 72 such headings) often reduce complex topics to objects of almost passing interest. For example, the discussion of communicative competence tests is contained in one paragraph; developments in language teaching are discussed in two paragraphs. A text on bilingual schooling cannot be expected to treat each topic covered with the same depth as that of the original sources. Yet such complex topicsfor example, the latest developments in second language teaching methodologydemand more than just cursory coverage. A careful synthesis is required, one that touches on primary issues and their implications for bilingual instruction. Such synthesis is absent from this volume, and the result is a choppy and awkward style that at best makes for tedious reading and at worst contributes to misinterpretation of the information presented. In many cases, no interpretation or definitions are offered for terms or concepts within the cited material, and no synthesis is offered to facilitate the readers understanding of the topic or why it relates to bilingual schooling. The poor organization of Bilingualism Through Schooling is evident throughout. The chapter summaries merely reflect lists in the chapter headings themselves. Little or no integration is evident in the volume, even though a great variety of research material has been summarized and categorized. Studies are cited, yet the author does not tie them together in an attempt to draw major implications. Tables are included to document conclusions of some studies, but many are simply impossible to interpret without more complete information. The compounding of meaningless detail in study citations and in incompletely explained tables is not helpful. Only rarely is there a synthesis of research that both illuminates a study and offers insight into larger issues. To compare how the two volumes treat important issues in the field of education and language minority students, I chose two researchers research groups that have made significant contributions to our understanding of both linguistic and cultural issues in bilingual schooling. One might expect a current work to include mention of the contributions of Skutnabb-Kangas, a prolific

researcher on European bilingualism and language minority issues. Her most recent volume, Bilingualism or Not (1983), is a substantive discussion of the issues related to minority and majority language students and the development of bilingualism in individuals and societies. Mention of her earlier work (Skutnabb-Kangas, 1979) and the importance of linking discussions of bilingualism in the United States with larger issues of minority-majority relations is not included in the Ramrez text. Ovando and Collier, on the other hand, cite her work in their discussion of the contexts in which first and second languages develop. It would be difficult to discuss instruction for students who are culturally and/or linguistically different from the mainstream without mentioning the work of the Kamehameha Education Research Institute (see Au & Jordan, 1981). The work of this group, which is central to discussions of the nature of cultural influences on learning styles and the classroom implications of these preferences, is discussed by Ovando and Collier, along with related research. By contrast, the Ramrez volume contains relatively little information on the relationship between language, culture, and education. Cultural information is covered in a chapter titled Attitudes Toward Language and Cultural Groups, which concerns itself with language attitudes. I also considered it important to compare the treatment of a particular topic that was included in both texts. The discussion of communicative competence demonstrated in microcosm the difficulties with the Ramrez text. Research studies are organized around the topic areas in a cut and paste format that does not provide context for the reader. In the Ovando and Collier volume, however, concepts are presented and synthesized; then research which illuminates the issues is examined. The development of a context for discussion in the Ovando and Collier text results in a clear exposition of the concept of communicative competence. The Ramrez text leaves much of the work of synthesis to the reader. The volume suffers, moreover, from a lack of balance. Some studies are briefly mentioned in terms of how they relate to the topic under discussion, but other studies (primarily those in which the author participated) are described in such minute detail that the motive for such a lengthy treatment and the relevance to general points are not always clear. Much of ones understanding of the Ramrez work depends on prior exposure to the topics discussed. As an assigned text for student reading, Bilingualism Through Schooling would require much in the way of interpretation. Beyond a beginning level, however, readers would be well advised to seek original sources to

get a more accurate picture of a particular research study. Thus, beginning-level students would find the lack of careful synthesis a serious drawback, while more advanced students would have already met with much of the material in its original form. Much of the difficulty in research and evaluation of bilingual programs stems from the false notion that bilingual education is a curriculum model rather than a curricular concept. Ovando and Collier, in Bilingual and ESL Classrooms, recognize the source of this difficulty. Thus, they stress the complexity of bilingual schooling and the need to consider and integrate a large number of variables, which may not always produce the same results for each local community or school. The integration of theory with the reality of teaching and learning in bilingual contexts is the strong point of this volume. While Bilingual and ESL Classrooms has the same goal as the Ramrez textto integrate substantial amounts of diverse theoretical and practical informationit proves much more successful in doing so. Its eight major chapters, each with several subsections, are well written, cohesive, and carefully edited. The first chapter, Students, addresses the nature of limited English proficient student populations and, in doing so, lets teachers and students speak for themselves. Subsequent chapters cover politics, programs, and resources; language; culture; social studies, music, and art; mathematics and science; assessment; and school and community. The practical realities of schooling, rather than theoretical topics, guide the organization of the text. A strong point of the text is the integration of concerns in both bilingual education and ESL. Historically, the relationship between the two fields has been one of mutual suspicion and animosity. While there is still strong belief on both sides about the need for L1 instruction, several important areas of study have recently emerged in which issues in bilingual education, foreign language education, and ESL overlap (Cummins, 1981, 1986; Saville-Troike, 1980; Wong Fillmore, 1982, 1986). Perhaps the rift will heal further as we continue to realize the complementary nature of our work and the need for our clients to be our primary focus. An attempt to bring together the perspectives and outcomes of all three areas in order to examine the academic needs of language minority students occurs in Bilingual and ESL Classrooms. An essential and often overlooked premise underlying bilingual education is the contribution of both language and culture to the instructional process. Bilingual and ESL Classrooms attends to the linguistic and sociolinguistic issues that might affect day-to-day classroom instruction. It very succinctly acquaints students with the

nature of language acquisition (first and second language learning for both children and adults) and the major theories contributing to our knowledge of those areas: The theories of Krashen and of Cummins, as well as the nature of bilingualism, are carefully outlined in a way that informs and amplifies. Language development in first and second languages is presented in a way that permits the reader to see the contribution of psycholinguistics, sociolinguistics, and second language acquisition research to the study of bilingualism. While the Ovando and Collier book is generally current in terms of both theory and resources, it is possible to fault the volume for the inclusion of information on federal programs concerning the support networks for bilingual instruction in the United States Evaluation, Dissemination, and Assessment Centers; Bilingual Education Service Centers; and so on. Generally such structures appear to change form approximately every 5 years as federal rules and regulations reorganize funding requirements. Inclusion of such inforrnation is usually fruitless, since it tends to be outdated almost immediately. Both books offer, at the end of each chapter, a list of recommended readings in specific topic areas. While Ramrez offers a short list of such sources, Ovando and Collier provide topicspecific recommendations for areas covered in the chapter. The quality and quantity of resources included in their book make it comprehensive, and sources of information are also given in several of the chapters. The reference section, too, is extensive and comprehensive. The suggested readings, which integrate theory and practice, almost preclude the requirement of a course bibliography, which has been such a necessary part of bilingual education foundations courses in the past. Clearly, the task of capturing and integrating the wealth of information needed to describe and discuss the reality of decisions concerning language minority students in the United States is a difficult one. Success at such a task can only enrich the resources for training teachers to work with language minority students. More important, it can serve to inform those outside the field of resources and options available for instructing bilingual students. Of these two current texts in the field of bilingual education, Bilingual and ESL Classroom can be recommended for successfully organizing a complex array of theory, research data, policy, and practice. Teacher education programs for bilingual education and English as a second language would find it an excellent primary introductory text. It might also be recommended for those

educators who want an understanding of the complexity of decisions to be made regarding teaching language minority children. REFERENCES
Anderson, T., & Boyer, M. (Eds.) (1970). Bilingual schooling in the United States. Washington, DC: U.S. Government Printing Office. Au, K. H., & Jordan, C. (1981). Teaching reading to Hawaiian children: Finding a culturally appropriate solution. In H.T. Trueba, G.P. Guthrie, & K.H. Au (Eds.), Culture and the bilingual classroom: Studies in classroom ethnography (pp. 139-152). Rowley, MA: Newbury House. Center for Applied Linguistics. (1977). Bilingual education: Current perspectives. Arlington, VA: Author. Cordasco, F. (1972). Bilingual schooling in the U. S.: A sourcebook for educational personnel. New York: McGraw-Hill. Cummins, J. (1981). The entry and exit fallacy in bilingual education. NABE Journal, 4 (3), 26-60, Cummins, J. (1986). Empowering minority students A framework for intervention. Harvard Educational Review, 56, 18-35. LaFontaine, H., Persky, B., & Golubchick, L. (1978). Bilingual education. Norwood, NJ: Ablex. Saville-Troike, M. (1980). Cross-cultural communication in the classroom. In J.E. Alatis (Ed.), Georgetown University Round Table on Languages and Linguistics 1980 (pp. 348-355). Washington, DC: Georgetown University Press. Skutnabb-Kangas, T. (1979). Language in the process of cultural assimilation and structural incorporation of linguistic minorities. Rosslyn, VA: National Clearinghouse for Bilingual Education. Skutnabb-Kangas, T. (1983). Bilingualism or not. Clevedon, England: Multilingual Matters. Trueba, H. T., & Barnett-Mizrahi, C. (1979). Bilingual multicultural education and the professional: From theory to practice. Rowley, MA: Newbury House. Wong Fillmore, L. (1982). Language minority students and school participation: What kind of English is needed? Journal of Education, 164, 143-156. Wong Fillmore, L. (1986). Research currents: Equity or excellence? Language Arts, 63, 474-480.

CONSTANCE WALKER University of Minnesota



Computational Linguistics: An Introduction Ralph Grishman. New York: Cambridge University Press, 1986. Pp. viii + 193. Computational linguistics is the study of computer systems for understanding and generating natural language. By understanding natural language generation in procedural terms, scientists can endow computer systems with the ability to generate and interpret natural language. This would make it possible for computers to (a) accomplish machine translation, (b) retrieve information from natural language texts, (c) engage in interactive communication with humans (e.g., for natural-language data-base retrieval with command language operations). These three classes of applications have been central in the development of computational linguistics, though the objectives of computational linguistics reach beyond these three goals. For example, with the assistance of computational linguistics, psychologists hope to build procedural models of the psychological processes underlying language acquisition and understanding, according to Grishman. Constructing a fluent, robust natural language interface is a difficult and complex task, notes Grishman (p. 7). To accomplish this task requires a thorough understanding of the roles that syntax, semantics, discourse, and information structuring play in the comprehension and generation of natural language. Reading and studying the chapters on these topics in Computational Linguistics will provide a framework for examining these roles. It will also familiarize the novice to the field with computational linguists work with several parsers. A parser scans an input line character by character in the attempt to derive a syntactic and grammatical meaning from the input. Christie and Christie (1984) state that a parser is a computer routine that controls decoding of an external program statement by establishing its syntactic tree, according to the specified syntax of the programming language (e.g., LISP). As Sippl and Sippl (1980) note, all compilers, assemblers, and interpreters use parsing routines to convert program source lines into a form the computer can understand. (According to Christie and Christie, a compiler is a program used to translate a programming language, e.g., BASIC or FORTRAN, into machine language that is understandable to a processor; an assembler converts the BASIC or FORTRAN into a form suitable for execution on a computer and provides error messages if needed; and an interpreter executes the BASIC or FORTRAN program by scanning, or parsing, each line of the

BASIC or FORTRAN code and changing it into machine code binary one and zeroes the computer understandseach time the program is run.) Grishman presents a brief description of several parsers used by linguists: context-free grammar parsers (e.g., the Harvard Predictive Analyzer), transformational parsers (e.g., Petricks parser), augmented context-free parsers (e.g., the New York University Linguistic String Parser), and context-sensitive parsers (e.g., General Electrics DEACON). The text also presents a discussion, albeit a rather cursory one, of the following: (a) the programming languages used to implement language analysis procedures (e.g., LISP and PROLOG); (b) the various types of logic (e.g., propositional and predictive) and the manner in which formal languages associated with these types of logic can be used to represent the meanings of sentences; and (c) the analysis of discourse, and knowledge organization and structuring (e.g., in terms of discourse frames, scripts, plans, and memory organization packets, or MOPS). Grishman provides, in other words, a conspectus of various issues dealing with the analysis and generation of natural language, issues which comprise the fulcrum of the investigations of computational linguists. Little background in computer science and finite mathematics is required for nascent computational linguists who will use the book as a course text or for curious readers from applied linguistics, psychology, or communication theory who pickup the book to gain insight into the field of computational linguistics. However, a background in transformational grammar will certainly help the reader understand the elucidation of sentential-level syntactic and semantic analyses. The chapter on discourse analysis and information structuring is the most reader friendly of all. It actually seems as though it was written with the nonlinguist in mind, but then it may be that discussion of scripts, information formats, and dialogue analysis contains language and content that is more accessible to the second language acquisition researcher, applied linguist, or cognitive psychologist. For those interested in obtaining in-depth information about the field of computational linguistics, Grishmans book will not suffice. According to the author himself, it will not provide the reader with a detailed survey of the grammatical theories and parsing algorithms used in computational linguistics (as does Winograds Language as a Cognitive Process: Vol. I. Syntax [1983]). Nor will it provide the reader with a large number of case studies and descriptions of specific systems to illustrate particular aspects of natural language processing (as does Tennants Natural Language Processing [1981]). However, it will provide the uninitiated with a

keyhole view of the major theoretical and procedural concerns of computational linguists. These linguists seek to understand natural language processes in order to create more natural computerhuman interaction. It is a valuable book for those interested in such interaction. REFERENCES
Christie, L. G., & Christie, J. (1984). The encyclopedia of microcomputer terminology: A sourcebook for business and professional people. Englewood Cliffs, NJ: Prentice-Hall. Sippl, C. J., & Sippl, R. J. (1980). Computer dictionary and handbook (3rd ed.). Indianapolis: Howard W. Sams. Tennant, H. (1981). Natural language processing. New York: Petrocelli. Winograd, T. (1983). Language as a cognitive process: Vol. I. Syntax. Reading, MA: Addison-Wesley. PATRICIA DUNKEL
The Pennsylvania State University

CAI Adaptation of Robert J. Dixsons Essential Idioms in English William B. Richardson and Sheldon Wise. New York: Regents/ ALA, 1985.

I Essential Idioms in English is a textbook with an accompanying

software package whose purpose is to teach 468 idioms to ESL students. The book and software (Apple II series; 48K minimum, DOS 3.3) may be used independently or in conjunction with each other. The three disks in the software package cover material corresponding closely to the 39 lessons in the book. Each of the 39 lessons on disk is further subdivided into four modules, or exercise types. Students need about 15 minutes to work a given module, so the entire package could involve students in 40 hours of interactive learning. The four modules that comprise each lesson on disk are called Learn the Idiom, Choose the Idiom, Write the Idiom, and Idiomatic. These modules drill students on a dozen idioms at a time in presentation, recognition, production, and quiz-game formats, respectively. This sequence of module presentation is subtly suggested; however, students may omit any of the modules or vary the order of presentation according to their individual learning styles by selecting from the menu.


The software also includes a Teachers Module, which allows teachers to activate or deactivate scoring, to invoke certain utility functions, and to set parameters which customize lesson delivery. Hence, this module is accessible to teachers but not to students. It is menu driven and transparent enough to allow easy experimentation (although the lessons are ready to run without resort to this module). Many practitioners of computer-assisted instruction (CAI) feel that the perceived threat to students of using computers to keep track of their scores outweighs any benefits derived; nevertheless, score keeping is one feature of the Teachers Module. This function is rather limited: Only 10 students may be registered, and a disk thus configured cannot be used by any but those 10 students unless the teacher turns the scoring off. Second, the Teachers Module may be used as a utility; for example, teachers can print (to screen or printer) student scores or lists of idioms used in any given lesson. Teachers may also conveniently preview the text employed in each module without actually working the problems in student mode. The third function of the Teachers Module is that it allows teachers to regulate how the lessons will be presented to the students. Features in this category include a quit option that can prevent students from leaving a module before completing it. In addition, the teacher can set the skip option so that when students choose to skip around difficult problems, these will reappear at the end of the lesson. The Teachers Module can also be configured so that students can turn off the sound effects. Unfortunately, these last options are of dubious value, unless testing is desired. The skip option is not very effective in learning mode because, with or without this option, students may skip around any items with the press of an arrow key; the skip option only ensures that these items will reappear (which, oddly, is not the default setting). Similarly, the quit option need not be set by the teacher, given that students working on their own ultimately have control over program exit via the on-off switch. Finally, since sound serves to motivate, distract, or even embarrass individual students, it seems inappropriate for teachers to infringe on the individuals right to turn it on or off. However, throughout these lessons, and no matter what the sound setting, the computer still makes a rasping sound each time there is an incorrect answer. I do not object to granting options to teachers, nor do I advocate imposing certain modes of using these lessons. However, on the basis of the hypothesis that computers are primarily learning (as opposed to teaching) devices, I believe that students should be given more options than teachers over their use of CAI. A prime

advantage of computers is that they can be adapted to individual learning styles through options available to students (Doughty, 1985; Jamieson & Chapelle, 1984; Stevens, 1984). Yet teachers are given the power here of overriding the student options purposely granted by the developers of this software if, through misconception or design, they choose to do so. Students using this software, unaware of the existence of a Teachers Module, are put directly into the lessons. Although they can choose from any of the modules, the first module in the suggested sequence is Learn the Idiom. The purpose of this module is to introduce the dozen idioms that will be drilled in each lesson in such a way that prior student preparation by teacher or text is not absolutely necessary. In each such module, the idioms are presented and defined, and examples of use given. Students can then examine the 12 idioms in or out of sequence, see the directions again (including an animated illustration of two-part verb separability), or quit this module by typing CTRL-Q. (However, if the quit option has been disallowed by the teacher, students must review all 12 idioms before they can exit.) On quitting this or any module, students see a histogram chart (in this case showing the number of idioms reviewed and the number skipped) before being prompted for their next activity choice. In the next module in the suggested sequence, Choose the Idiom, students select from triads of idioms those that will fit the blanks in a series of sentences. They can experiment, trying out any or all the phrases in the blanks, and press RETURN when the sentence looks right. At this point, the phrase literally falls out of the sentence and is either caught by a little man, who tosses the idiom back into the sentence and announces Correct! or the phrase disintegrates into dust, and the little man comes out with a broom and rapidly sweeps the piles of dust away. Here, the computer is used to make feedback catchy and immediate in a unique and appropriate application of computers to language learning. Once students have studied the idioms and completed the recognition exercise, they have a chance, with Write the Idiom, to produce the idioms from recall. Here again, the computer is used to good advantage providing progressively more feedback as students need it. Initially, there is no feedback; students are confronted with a continuous blank line in the middle of a sentence, in which they are to type one of the idioms they have learned. But on each subsequent attempt at the problem, more information is given. First, the continuous line becomes as many blanks as there are letters in the answer. Then, correct letters are left in place, and


students can see a list of idioms practiced in that lesson. Eventually, students are guided to the answer, or they can skip to the next problem. The final lesson module, Idio-matic, invites students to race a clock in determining which of three idioms fits in a blank in a given sentence. As sentences appear on the screen, students can toggle the three possible answers, one after another, into the blank until they think one is correct. Students may vary the speed; the faster the game, the higher the point value, but the faster the points are ticked off during periods of indecision. The exercise is challenging and fast paced; some students become truly excited while attempting to rack up points at this game. This exercise is another which appropriately exploits the medium. These lessons take good advantage of many aspects of computers which make them uniquely suited to language learning. On the other hand, the lessons are not immune from what might be called the tyranny of computers. One example is the Title Page routine, which students are presented with from time to time, for example, on exiting one of the four modules. While entertaining at first, this routine eventually becomes tedious because a quick and easy exit is neither provided nor clearly noted on the screen. Scoring also imposes a subtle tyranny on these lessons, apart from the test-like rigidity possible with the quit and skip options already noted. For instance, as students exit the Choose the Idiom module and see the histogram telling them how they have done, they find that even if they have worked constructively with the computer to get a problem right at the second or third attempt, all problems not answered correctly on the first try have been counted wrong. This decidedly negative feedback and lack of credit for effort could cancel out student willingness to use the program in the cooperative way its designers intended. A third tyranny is imposed by the medium itself, by the fact that there is only so much programming that can be put on a floppy disk. This constraint is most obvious in Learn the Idiom, in which the intent is clearly to introduce the idioms independently of any other material, but idiom definitions sometimes seem incomplete or enigmatic, and there is little interaction with students. The fact that this particular module takes so little advantage of the medium over what is possible with a book is in part a consequence of limitations of memory and disk space (Sheldon Wise, personal communication, July 10, 1985), In this event, I am concerned that memory space which could have otherwise been devoted to enhancing lesson interactivity has been allocated to options of limited value to teachers and of almost no value to students. I believe this to be a

serious issue, as it suggests conflicting perceptions of educational software from marketing and pedagogical perspectives. A final problem encountered with all Regents/ALA materials is that despite their consistently high quality relative to competitive offerings in the field, the programs are rendered difficult to work with due to strict backup and copy protection policies. This touches on another controversy in commercial educational software: At what point does an institutions need to duplicate software for inhouse use infringe on the rights of authors to profit from their efforts? Institutions need to duplicate software to preserve masters and to furnish copies in numbers needed by students working at individual stations. Regents/ALA will provide backup copies at reduced prices, but only if these are purchased at the same time as the original courseware. As for multiple student copies, the position of the publishers is made clear in the documentation that accompanies this packet: Multiple copies may not be made under any circumstances. Additional diskettes may be purchased from Regents/ALA. The diskettes are protected by a sophisticated anti-duplication system; any unlawful attempts to duplicate them may damage the diskettes, voiding the warranty and licensing agreement. (p. 2) This policy makes it an expensive proposition to use this packet lawfully in any educational setting other than as part of a library of materials available to students on a check-out basis. Despite its inevitable deficiencies, I find this to be a respectable offering in the area of computer-assisted language learning and one worthy of high recommendation. The integrity of lesson content is high; that is, the exercises are generally meaningful and well contextualized, instructions are clear, and the text is devoid of typographical errors. The lessons are motivating, make imaginative and proper use of features inherent in the medium (e. g., immediate and constructive feedback; appropriate use of graphics and animation), and use language and methodology suitable to ESL. Furthermore, the package is highly professional in its programming and presentation and apparently has been thoroughly tested. (I found only one bug: If the program is aborted prematurely, the screen fails to clear, and Goodbye is written over the lesson menu on the screen. Since the program runs normally after that, this bug is of little consequence. ) The shortcomings mentioned regard departures from my conception of an ideal software packet, and my remarks neither address marketing considerations nor deal thoroughly with limitations of memory and disk space, which

constrained the developers of this software and which are a major, if invisible, consideration in any CAI implementation. The quality of this software is also high relative to other commercial software currently available for ESL which teaches discrete skills.1 One such program (Layton, 1985) requires students to contrive grammatically correct English sentences in response to rebus-like formulas, a practice weighted more toward tedium than toward communication. Another program for ESL (Hamilton & Hombs, 1985) combines stick-like figure graphics with drill-book exercises. This combination, its producers claim, merges left- and right-brain activities in integrated learning. Such programs support Daiutes (1985) remark concerning commercial software: that when teachers see the available programs . . . they are often disappointed (p. 15). In contrast, Essential Idioms is motivating, challenging, relevant, and imaginatively programmed. It is not yet common for textbooks to be accompanied by software; more often they are paired with workbooks, audiocassettes, or video materials. But as computers continue to proliferate and as more and more constructive uses are discovered for them in language learning, it seems logical that an increasing number of textbook authors will consider the possibilities of supplementing their work with courseware, either to serve as interactive workbooks, along the lines of the software reviewed here, or in even more imaginative capacities. Looked at in this perspective, Essential Idioms is an innovative and welcome prototype. REFERENCES
Baltra, A. (1984). An EFL classroom in a mystery house. T E S O L Newsletter, 18 (6), 15. Biggie, L. (1984). Public domain software. TESOL Newsletter, 18 (3), 11. Daiute, C. (1985). Writing and computers. Reading, MA: Addison-Wesley. Doughty, C. (1985, April). Computer-assisted instruction based on individual learner strategies. Paper presented at the 19th Annual TESOL Convention, New York. Dutra, I. (1985, April). Hypothesis testing and problem-solving software for ESL students. Paper presented at the 19th Annual TESOL Convention, New York.
1 Many

programs exist for ESL besides commercial software teaching discrete skills. Examples of more communicative programs are detailed in Higgins and Johns (1984); Underwood (1984); Ferreira, Sklar, and Kagan (1984); Baltra (1984); Dutra (1985); and Johnson (1987). In addition, numerous programs in the public domain are suitable for ESL, as reported in Biggie (1984) and Stevens (1985).



Ferreira, L., Sklar, S., & Kagan, A. (1984, March). Computers as realia: Using existing software to develop ESL communication skills. P a p e r presented at the 18th Annual TESOL Convention, Houston. Hamilton, P., & Hombs, B. (1985). Core reading and vocabulary development. Freeport, NY: Educational Activities. Higgins, J., & Johns, T. (1984). Computers in language learning. Reading, MA: Addison-Wesley. Jamieson, J., & Chapelle, C. (1984, March). Computer assisted language learning: Is it for everyone? Paper presented at the 18th Annual TESOL Convention, Houston. Johnson, N. (1987). Task-based CALL activities. C.A.L.L. Digest, 3 (2), 12 . Layton, T. (1985). ESL picture grammar. New York: Gessler Educational Software. Stevens, V. (1984). Implications of research and theory concerning the influence of choice and control on the effectiveness of CALL. CALICO Journal, 2(l), 28-33,48. Stevens, V. (1985). Youd be surprised at how much public domain software you can adapt to ESL and language learning. TESL Reporter, 18 (l), 8-15. Underwood, J.H. (1984). Linguistics, computers, and the language teacher: A communicative approach. Rowley, MA: Newbury House.

VANCE STEVENS Sultan Qaboos University, Oman




The TESOL Quarterly invites readers to submit short reports and updates on their work. These summaries may address any areas of interest to Quarterly readers. Authors addresses are printed with these reports to enable interested readers to contact the authors for more details. Edited by D. SCOTT ENRIGHT
Georgia State University

Cognitive Style and First Language Background in Second Language Test Performance
Brigham Young UniversityHawaii Campus

The influence of adult second language learners first language background on language test performance has received recent attention in the testing literature. Different group performance profiles emerge on language test batteries, and this suggests that language background should be incorporated into a redefinition of language proficiency (Brutten, Mouw, & Perkins, 1986; Farhady, 1982; L. Hansen, 1984a). Attention has also been called to the need for awareness of possible group bias in test selection and interpretation (L. Hansen, 1984b). The influence of test takers cultural background has been shown to be a factor in performance on particular types of language test. On dictations, for example, Filipino and South Pacific island students were found to perform significantly better than EFL Asian groups; at the same time, no significant differences in the performance of the groups emerged on most of the other measures in an ESL test battery (Evans & L. Hansen, 1986). Test profiles of these groups suggest that the island students, from ESL backgrounds in cultures characterized by strong oral traditions, may tend to score higher on tests which involve spoken English, while the performance of the EFL Asians may be relatively better on tests limited to the written word (L. Hansen, 1984a). The two groups also differed in performance on cloze, although at a lower level of significance than on the dictation (Evans & L. Hansen, 1986). Other research suggests that cognitive style may contribute to such variation between groups (L. Hansen, 1984b; Stansfield & J. Hansen, 1983); quite apart from their proficiency in the language being tested, individuals with a field-independent cognitive style appear to have a slight advantage over those with a field-dependent style. In multicultural groups, therefore, a bias in the cloze testing technique may work to the disadvantage of more field-dependent individuals or groups.

The study reported here sought to discover further evidence of cultural differences in cognitive styles and of the effects of these differences on second language test performance by examining the following questions: (a) Are there significant differences between South Pacific and EFL and ESL Asian groups in their levels of field dependence-independence (FD/I) and in their reflectivity-impulsivity (R/I)? and (b) How do the two cognitive styles relate to the language test performance of the students? METHOD Subjects were enrolled in the English Language Institute (ELI) at Brigham Young UniversityHawaii Campus. The 108 female and 101 male subjects were divided into three demographic groups: (a) 63 South Pacific (Tonga 28, Samoa 20, Micronesia 15); (b) 77 EFL Asian (Japan 26, Korea 28, Republic of China 8, Peoples Republic of China 6, Indonesia 5, Thailand 3, Vietnam 1); and (c) 69 ESL Asian (Hong Kong 53, Philippines 16). The measure of FD/I, the Group Embedded Figures Test (GEFT) (Witkin & Goodenough, 1981), was administered to the students in their ELI classrooms. The GEFT is an 18-point test on which a high score indicates a high level of field independence. The measure of cognitive tempo, the Matching Familiar Figures Test (MFFT) (Messer, 1976), was administered individually to a randomly selected subgroup of 53 students from the larger group: 18 South Pacific islanders, 20 EFL Asians, and 15 ESL Asians. Two measures were obtained from this match-to-standard test: (a) response latency, that is, the time between presentation of the item and the subjects first response, and (b) response accuracy, the number of errors. A double-median split procedure was employed for the classification of conceptual tempo: Those scoring above the median on response latency and below the median on errors were classified as reflective (slow/accurate); those scoring below the median on response latency and above the median on errors were classified as impulsive (fast/ inaccurate). The ELI final examination battery included (a) the Michigan Test of English Language Proficiency (MTELP); (b) the Michigan Test of Aural Comprehension (MTAC); (c) an essay which was graded by two readers on a 10-point scale; (d) a dictation, read three times, the second with pauses for writing; (e) a cloze passage with every seventh word deleted and scored by the acceptable-synonym method; and (f) a speaking test consisting of subtests for conversation, reading pronunciation, and a prepared speech. These three subtests were administered on three consecutive days, and each was rated by a different pair of examiners. RESULTS AND DISCUSSION The mean scores on the GEFT were generally higher for the Asians than for the island students (Table 1). To determine the statistical significance of the differences, a two-way analysis of variance (ANOVA) was run, in

which the data were partitioned into three groups as described above. The results, presented in Table 2, show that the F value for culture is significant at the .001 level, and the F value for sex at the .01 level. Thus, FD/I cognitive style was associated with cultural background as well as sex. The Asians tended to be more FI than the South Pacific islanders, and the male students more than the female.
TABLE 1 GEFT Descriptive Statistics by Culture and Sex

Culture Samoan Micronesia Filipino Other Asian Korean Hong Kong Japanese


Male M

Female M

TABLE 2 Two-Way ANOVA for Culture and sex With GEFT as Dependent Variable Source SS df MS F

* p <.01.

** p <.001.

TABLE 3 MFFT Mean Response Latency Times and Error Totals by Cultural Groupings

The mean response latency times (in minutes) and error totals, as measured by the MFFT, are presented in Table 3. The statistical significance of group differences in conceptual tempo was determined by two-way ANOVAs, which were run for latency time and for error. The results, given in Table 4, show that for time, the F value for culture is

significant at the .01 level, while that for sex is not significant. Similarly, the cultures differed significantly in number of errors, at a level of .001, again with no sex differences. Thus, it appears that the cognitive tempo of these ESL students is associated with cultural background. The South Pacific islanders tended to be more impulsive, the Asians more reflective.
TABLE 4 Two-Way ANOVAs for Culture and Sex With MFFT Latency Time and Error

The second question examined in the study concerned the relationship between the cognitive variables and second language test performance. Correlational procedures (Pearson product-moment) were therefore used to determine the direction and strength of the relationship between scores on the cognitive measures and the English test scores (see Table 5).
TABLE 5 Correlations Between GEFT, MFFT Time and Errors and the Language Test Scores

All of the GEFT correlations, though of low magnitude, are positive, indicating that a higher level of field independence tends to be associated with higher scores on language tests. The slightly higher correlation of the GEFT with the cloze test than with the other tests seems to support previous findings of a small bias in favor of field-independent learners on this test format. Cognitive tempo, on the other hand, showed no apparent relationship with performance on the language tests, as seen in the nonsignificant correlations between the test scores and both MFFT time and errors. Correlational research has limitations, however, in the search for effects of learner characteristics on such a complex phenomenon as language test performance. Because so many factors interact to influence the testing process, some of these factors may not have a consistent and statistically obvious relationship to language test scores for all learners and in all testing situations. To achieve a useful understanding of the consequences of various cognitive-style profiles for language test performance, research may need to go beyond correlational studies which look only for a simple main effect. Factorial research designs, for example, would facilitate the sorting out of interaction effects. REFERENCES
Brutten, S. R., Mouw, J. T., & Perkins, K. (1986). The effects of language group, proficiency level, and instruction on ESL subjects control of the {D} and {Z} morphemes. TESOL Quarterly, 20, 553-559. Evans, N., & Hansen, L. (1986, March). Dictation: A test of what? Paper presented at the 20th Annual TESOL Convention, Anaheim, CA. Farhady, H. (1982). Measures of language proficiency from the learners perspective. TESOL Quarterly, 16, 43-59. Hansen, L. (1984a). The ESL noise test: Cultural differences in affect and performance. In P. Larson, E.L. Judd, & D.S, Messerschmitt (Eds.), On TESOL 84 (pp. 55-61). Washington, DC: TESOL. Hansen, L. (1984b). Field dependence-independence and language testing: Evidence from six Pacific island cultures. TESOL Quarterly, 18, 311-324. Messer, S.B. (1976). Reflection-impulsivity: A review.Psychological Bulletin, 83, 1026-1052. Stansfield, C., & Hansen, J. (1983). Field dependence-independence as a variable in second language cloze test performance. TESOL Quarterly, 17, 29-38. Witkin, H. A., & Goodenough, D.R. (1981). Cognitive styles: Essence and origins. New York: International Universities Press.

Authors Address: Box 1870, Brigham Young UniversityHawaii Campus, Laie, HI 96762-1294



An Overview of Undergraduate ESL Program Models: A Comparison of Administrative Policies for International Students
Arizona State University

The focus of most research in academic ESL has been on methodology and applied linguisticsthe classroom and the textbook. Little attention has been paid to the larger issue of how ESL programs as a whole serve the needs of the university and the international student. Thus, ESL teachers or administrators attempting to gain an overview of their own local program in regard to national trends have been hampered by the inaccessibility of data in this area. To help overcome this obstacle to effective program evaluation and innovation, a survey of ESL programs in higher education was undertaken to explore the following issues:
1. What are the features of ESL programs in higher education, and how do

the programs compare with one another in seeking to meet institutional and student needs? 2. Can similar . programs be grouped to form representative constructs or . models? Can these models demonstrate how-the programs function in regard to admissions (i.e., TOEFL score requirements), course options, and the coordination of intensive English classes with freshman English courses?

MODELS OF ESL PROGRAMS A survey of ESL programs at 28 American universities, including those in the 10-member Pacific Athletic Conference, the Big Ten, and the Big Eight Conferences, was conducted in 1985 and updated in the fall of 1986 and the spring of 1987. Since it was not possible to conduct a large-scale survey of the majority of universities in the nation, these 28 were chosen as peer institutions whose policies were of interest to the administration of Arizona State University. When information relating to the questions above was tabulated, it was found that the majority of the 28 ESL freshman English programs studied could be classified according to four models: the conservative model, the traditional model, the bridging model, and the progressive model. Northwestern University, the University of Oklahoma, and Stanford University were not included in the classification, as these universities have no special program for undergraduate international students, who are mainstreamed with native English-speaking students.

Conservative Model Conservative commonly means tending to preserve existing conditions or resisting change. The conservative model does, in fact, maintain the same program for ESL students as that designed for native Englishspeaking students, with minimal modifications to accommodate the special needs of international students (see Figure 1). In this model, the simplest of the four, there are only two options. Students who have not attained the required TOEFL score take intensive English classes for nonadmitted students, while those with the required TOEFL score take a credit-bearing freshman English class for international students that is supposed to parallel the standard freshman English course for native speakers. These students meanwhile carry full academic loads. There is no on-site post-TOEFL testing of international students and no bridging program for admitted students. Arizona State University and Oklahoma State University follow this model. FIGURE 1 Conservative Model TOEFL

Intensive English ( noncredit )

Note: The programs of Arizona State University and Oklahoma State University (nonprofessional students) conformed to this model. Traditional Model Traditional usually means following customs or conventions previously established. The traditional model has followed the tradition established for native speakers with weak skills by adding remedial writing courses to the curriculum (see Figure 2). In this model, the three main options are (a)

full-time intensive English courses for nonadmitted students, (b) a prefreshman English course or courses for conditionally admitted students or for regularly admitted students with low scores on the on-site placement test, and (c) a credit-bearing freshman English course for admitted students with higher scores on the placement exam. Students with exceptionally high scores may be exempt from freshman English.

FIGURE 2 Traditional Model

Intensive English


) .

Note: At the time of the survey, the programs of the University of Arizona, the University of California, Berkeley, the University of California, Los Angeles, Iowa State University, Kansas State University, The Ohio State University, the University of Oregon, and Washington State University conformed to this model. Recently, Purdue University has instituted a credit-bearing, prefreshman English course for international students.

At the time of the survey, 3 universities offered a prefreshman English course for conditionally admitted students, while 5 institutions provided a prefreshman English course for low-scoring admitted students. Purdue University has recently instituted a two-option variation on this model. While there is no intensive English program at Purdue and no on-site testing, undergraduate international students in the freshman English course who are found to have poor English skills are transferred into a credit-bearing, prefreshman English course by teacher recommendation.

Bridging Model Bridging usually means providing a connection or transition between two points. The bridging model provides a transition between the intensive English classes and freshman English (see Figure 3). It is the most popular model and has been adopted by 12 universities. While there are again three main tracks, additional options are offered within these tracks. The first track is the intensive English program for nonadmitted students, the second is a series of semi-intensive English classes (either credit or noncredit), and the third is freshman English.

FIGURE 3 Bridging Model

Note Of the programs which conformed to this model, those at the University of Colorado, the University of Illinois, Indiana University, the University of Iowa, the University of Michigan, the University of Nebraska, Oregon State University, and the University of Wisconsin offer credit for semi-intensive English courses; those at Michigan State University, the University of Minnesota, the University of Missouri, and the University of Washington do not. The University of Michigan program previously fit this model; recently, however, the intensive English classes have been eliminated.

An intensive English program is here defined as a full-time program of English classes teaching all four skills on varying levels; students attend approximately 20 hours of classes a week. A semi-intensive tract is a parttime program in which conditionally admitted or regularly admitted students with low scores on the on-site placement test may spend from 515 hours a week in English classes. Instruction on each of the four skills (reading, writing, speaking, and listening) is presented in separate classes, but students often take only those skill classes in which they are weakest, as determined by on-site testing. Freshman English is usually a one- or

two-semester course emphasizing essay writing and a research paper. This course usually meets 3 hours a week and carries 3 credits a semester. The bridging model differs from the traditional model, in which intermediary courses usually correspond to traditional remedial English courses for native speakers. These courses can be classified as prefreshman English, since they concentrate on writing rather than on the four major skills included in ESL courses in the bridging model. The schools adopting the bridging model can be further divided into those that grant credit for these semi-intensive English courses (8) and those that do not (4). A further option at many of the universities using the bridging model is admission to the university upon completion of the intensive English program, in lieu of the TOEFL requirement. Usually these students must be recommended by the ESL department and must have a TOEFL score close to the university TOEFL cutoff score. Progressive Model
Progressive generally means favoring progress, change, improvement, or reform as opposed to wanting to preserve things as they are. The progressive model has eliminated the common barrier of the TOEFL exam for entrance to the university, allowing students to ease into full-time university work as English skills improve (see Figure 4). This model has been adopted by 2 of the schools in the surveythe University of Kansas and the University of Southern California.
FIGURE 4 Progressive Model

Note: The programs of the University of Kansas (in which no credit is offered for intensive English) and of the University of Southern California (in which up to 12 credits may be earned in intensive English) conformed to this model. 574 TESOL QUARTERLY

In this model an open undergraduate admissions policy is followed: Students are not barred from university admission by failure to meet or exceed a TOEFL cutoff score; they are admitted on the basis of their previous academic records. Again, there is a three-track program for the admitted students, depending on the results of the placement exam: (a) Those with low scores go into a full-time intensive English program; (b) those with middle scores (the majority of students) go into semi-intensive English courses, with a reduced load of university classes; and (c) those with high scores go into freshman English classes. DISCUSSION The four models above can be compared and evaluated on the basis of how well they serve the needs of both the university and the international student. Probably the weakest of the models is the original one developed for ESL in the university, the conservative model. This model has been called the sink or swim model by Hargett and Olswag (1984), since the student is given little help in making the adjustment from full-time language classes to full-time university classes. It is assumed that after students have attained a specified TOEFL score, their language abilities are ,then fully adequate for university work, and students are given a load and schedule identical to that of a native speaker. In fact, as the majority of ESL administrators and teachers contacted for this survey pointed out, a fairly high TOEFL score does not guarantee academic success. A score of 500 or even 550 does not ensure that students will have the speaking, writing, and note-taking skills necessary to carry on university work (Huckin & Olsen, 1984). The freshman English course for international students in the conservative model, parallel in objectives to the native-speaker course, would seem to exact high standards from the students immediately. In fact, because of the diverse skill levels of the class members, the goals of the course are often difficult to meet (Gaffney & Mason, 1983), and international students will almost certainly exit without skills comparable with those of native speakers completing a regular freshman English course. The traditional model, with prefreshman English as a transition between intensive English classes and freshman English, still has certain drawbacks, although it is superior to the conservative model. According to Pearson (1981), students supposedly ready to begin university work need to read with speed and comprehension, to write cogent essays and reports, to understand and take notes from lectures, and to employ effective study techniques (p. 413). Studies by Johns (1981) and Ostler (1980) also point up the need for listening and reading skills. Thus, a major weakness of the rigid 3-credit, 3-hour writing course in the traditional model is that it does not attack deficiencies in speaking, reading, and note-taking skills essential for academic success. Perhaps the most flexible model is the bridging model, which allows students with low scores on the placement test as well as conditionally

admitted students to take a series of courses to bring their academic English skills up to a level at which they can succeed in their university classes. Since students often carry a reduced academic load while taking the semi-intensive classes, they are allowed a period of transition. This model differs from the traditional model in that the semi-intensive English courses are usually taught by ESL specialists familiar with the problems of international students, often faculty from the intensive English program. Unlike the traditional model, the bridging model develops not only writing but other skills as well. The University of Washington, for example, has a two-track listening/reading-writing program, and students usually test into one of these two tracks. At Michigan State University, students may take several possible courses (grammar, listening/speaking, language lab, reading, and writing). The advantage of the bridging model over the progressive model is that there is still control over admissions. Most of the ESL specialists contacted seemed to feel that on-site testing addressed limitations of the TOEFL but that the TOEFL, as a standardized test available all over the world, served a definite purpose. However, the innovative progressive model offers special opportunities for the international student that are not available with the other models. Since students in universities using the progressive model are admitted to the university on the basis of their academic records, they are not held back by their present English ability. Also, students in these programs are given a university schedule geared specifically to their needs and may immediately begin to take one or two university courses such as mathematics, if they place in the high-intermediate or advanced level. These characteristics of the progressive model make it appealing in three ways: (a) It can be considered more democratic, since English language instruction programs are only available to the upper classes in many countries (Kaplan, 1968); (b) it dispenses with problems and complaints sometimes leveled at the TOEFL, such as language bias (Farhady, 1979; Hosley, 1978) and cultural bias (Traynor, 1985); and (c)it relies on the assumption that academic success in the university depends more on previous academic success than present English ability (Kaplan, 1968; Valdes, 1977). The main administrative drawbacks of this model may be the pressure put on the ESL program to bring all students up to a very high level of English proficiency in a short time and the problem of eliminating those students who, for one reason or another, are unable to attain this level. CONCLUSION The results of the survey reported here suggest that the 28 programs which participated have clear-cut differences which generally can be characterized by four distinct models. This finding points to the conclusion that while methodology and linguistic theory are legitimate concerns for ESL teachers and administrators, attention must also be paid to the overall ESL program format.

Most ESL educators would agree that the goal of the ESL program in higher education is to produce students whose listening, speaking, reading, and writing skills in English allow them to succeed in their university classes and, in Ostlers (1980) terms, to compete on an equal basis with American students (p. 501). The ability to reach such a goal and to produce such students stretches beyond the classroom to the ESL program as a whole. It is hoped that this overview of undergraduate ESL program models will provide a preliminary framework to assist in the evaluation and improvement of ESL programs. It is also hoped that other studies outside of the traditional methodological and linguistic boundaries will be undertaken to point to new directions for the development of ESL in higher education.
REFERENCES Farhady, H. (1979). Test bias in language placement examinations. In C. Yorio & J. Schachter (Eds.), On TESOL 79 (pp. 162-170). Washington, DC: TESOL. Gaffney, J., & Mason, V. (1983). Rationalizing placement and promotion decisions in a major ELT program. TESOL Quarterly, 17, 97-108. Hargett, G. R., & Olswag, S.G. (1984). An institutional approach to improving English proficiency of foreign students: The modified transitional model. The American Language Journal, 2(l), 67-83. Hosley, D. (1978). Performance differences of foreign students on the TOEFL. TESOL Quarterly, 13,209-217. Huckin, T. N., & Olsen, L.A. (1984). The need for professionally oriented ESL instruction in the United States. TESOL Quarterly, 18, 273-294. Johns, A. (1981). Necessary English: A faculty survey. TESOL Quarterly, 15, 51-57. Kaplan, R. (1968, Winter). Teaching English and international exchange. Exchange, pp. 43-47. Washington, DC: U.S. Information Agency, U.S. Advisory Commission on International Education and Cultural Affairs. Ostler, S.E. (1980). A survey of academic needs for advanced ESL. TESOL Quarterly, 14, 489-502. Pearson, C.R. (1981). Advanced academic skills in the low-level ESL class. TESOL Quarterly, 15, 413-423. Traynor, R. (1985). The TOEFL: An appraisal. ELT Journal, 39, 43-47. Valdes, J. (1977). Assessing English proficiency in the foreign student admission process. College and University, 52, 521-528.

Authors Address: American Language and Culture Program, Arizona State University, Tempe, AZ 85287



The D. C. Schools Project

MARTHA FARMELO Center for Immigration Policy and Refugee Assistance Georgetown University Since January 1984, the D.C. Schools Project of the Center for Immigration Policy and Refugee Assistance (CIPRA) at Georgetown University has developed a pool of approximately 200 volunteers who help non-English-speaking children and adults enrolled in D.C. public schools to learn English as a second language and adjust to North American life. The volunteers work as tutors in several local public schools, on Georgetowns campus on Saturdays, and in the homes of recently arrived families. The project serves as an adjunct to the Division of Bilingual Education of the D.C. Public School System and, because of its effectiveness, is recognized as an official volunteer program of the D.C. Public School System. The project was initiated in response to a media report describing how a large influx of non-English-speaking students was overwhelming the District of Columbia Public School System (Guillermoprieto, 1983). Most of these students arrive with little or no formal education and often suffer the psychological trauma associated with violence in their native country, the harrowing experience of migration to the United States, and indefinite separation from family members. Moreover, oftentimes financial problems not only force parents to work two full-time jobs but also pressure the students into entering the work force at a young age. The decision to initiate a tutorial program was based on (a) the immigrant students immediate need for survival English language skills, as well as general academic assistance, and (b) the teachers and administrators need for continual reinforcement of their efforts to retain and educate these students with unique educational and developmental needs. The principal roles of the volunteers in the D.C. Schools Project are (a) to provide students with the individualized attention they are not afforded in their crowded classrooms, (b) to act as a liaison between the students and families and the various human service agencies in the community, and (c) to serve as role models and friends to students. Volunteers are encouraged to develop relationships that extend beyond the tutoring setting and in this way to assist in the students personal as well as academic development. Through the encouragement and support of their tutors, the students gain confidence in themselves and their abilities and become more aware of the importance of learning English and excelling in school in order to better their future educational and employment opportunities. PROGRAMS The D.C. Schools Project actually consists of four youth tutoring programs (In-School, One-to-One, Saturday Morning, Peer-Tutoring) and

two adult tutoring programs (Gordon Adult Education Center, Employee Education). For the In-School program, volunteers go to the schools once or twice a week, Monday through Friday, in 1 l/2-hour segments, either at 9:30 a.m. or 1:30 p.m. Transportation is provided from Georgetown University. Working under the supervision of the ESL teachers, volunteers tutor one or two ESL students in basic or intermediate English. Because volunteers also seek to establish themselves as friends and role models to the students, each tutor usually works with the same student(s) each time. Tutors are encouraged to meet the students family and to participate in group outings organized by the CIPRA staff or other project volunteers. The volunteers are part of the school systems Operation Rescue and Operation Outreach programs, which are designed to assist all elementary and secondary students. In the One-to-One program, volunteers visit students homes for an hour and a half twice a week for individual tutoring. In addition to trying to establish themselves as role models and friends to the children, volunteers also attempt to serve as resource persons for the family. Volunteers are encouraged to meet with their students outside of the regular tutoring time and environment. Transportation is provided from the Georgetown campus, Monday through Thursday evenings. The Saturday Morning program consists of volunteers who, working under the supervision of the CIPRA staff, tutor non-English-speaking junior high school students on the Georgetown campus. In the morning, they provide the students with one-on-one assistance in speaking and writing English; at noon, volunteers and students eat lunch in a large group; and in the afternoon, tutors and students visit museums, play games, or partake in other activities. In the Peer-Tutoring program, junior high school students tutor elementary-age peers in basic English language skills on 2 afternoons a week from 3:30 p.m. to 4:30 p.m. The peer tutors, who have taken part in the activities of the D.C. Schools Project and who have excelled in their ESL program, work under the supervision of the Program Coordinators. Through preprogram training and one-on-one tutoring, the tutors as well as the students refine these language skills. The tutors have the opportunity to establish themselves as role models and develop leadership skills. In the Gordon Adult Education Center program, volunteers, working under the supervision of the Centers Volunteer Coordinator, serve as teacher assistants in the ESL classroom. (The Gordon Adult Education Center is administered by the D.C. Public School System.) These students are preparing for college placement exams or are learning basic skills necessary for job placement. The Employee Education program involves volunteers providing basic English instruction twice a week to non-English-speaking Georgetown University employees. This program functions Monday through Friday from roughly 3:00 p.m. to 4:00 p.m.

STAFFING AND TRAINING The majority of the volunteers are undergraduate students at Georgetown, and the balance is made up of alumni of the University, graduate students, young professionals in the D.C. area, and students from local senior high schools. Volunteers are recruited by (a) word of mouth (as current or recent students, the Project and Program Coordinators maintain wide contacts at the University), (b) flyers, (c) presentations in classes, (d) project participation in the annual Student Activities Fair, as well as information and recruitment tables in the student center and outside of the library and the cafeteria at the beginning of each semester, and (e) contacts with local high school counselors and teachers who help to enlist students for the Saturday Morning program. In anticipation of a lack of Georgetown University student volunteers during the summer periods, volunteers have been recruited for these periods by (a) a letter sent to over 2,000 area University alumni each May asking them to volunteer for the summer and then encouraging them to continue during the academic year, (b) a similar letter sent to current undergraduate students whose permanent address is in the Washington area, (c) posters and notices in University publications, and (d) word of mouth. In a given year, over 300 volunteers participate in the various programs, assisting well over 500 limited English proficient students. In the One-toOne program, the largest each year, approximately 150 volunteers tutor over 300 students. Approximately 80 volunteers tutor the same number of students in the Saturday program, and about 60 volunteers tutor approximately 100 students in the In-School program. In addition, about 20 peer tutors work with an equal number of students in that program, and 10 volunteers tutor at the Gordon Adult Education Center and about the same number in the Employee Education program. The exact number of students being tutored is difficult to record, since volunteers in the One-to-One program are assigned to work with particular students but often find themselves working with several other family members and since volunteers in the In-School and Gordon Center programs may work with individual students, serve as teachers assistants in ESL classes, or rotate among small groups of students, all according to the needs of the individual teachers. Each volunteer participates in a four-step training and orientation program consisting of an individual orientation, training seminar, volunteer dinner, and individualized follow-up. In the first step, one of the Coordinators discusses the project for approximately 30 minutes with each volunteer before the volunteer begins work. The discussion covers the history of the program, including its relationship with the D.C. Public School System, the continued demonstrated need for the service to be provided by the volunteers, experiences of other volunteers, and the particular aspects of the specific program in which they are to participate. Through both the distribution of

a booklet on human services available in the Adams Morgan neighborhood and discussion with the Program Coordinators, volunteers are also made aware of the services provided by the school system and community to the students and their families. Several times during the semester, a 2- hour training session on ESL tutoring is provided by Education Specialists from the Bilingual Education Division, the Office of Volunteer Services and Training of the D.C. public schools, and ESL teachers from the various schools that are serviced. Approximately 20 to 30 volunteers attend each session. During every semester, each volunteer is encouraged to attend a dinner with eight or nine other volunteers to discuss issues and problems associated with the work. On occasion, representatives of the Division of Bilingual Education or ESL teachers are present at these dinners. CIPRA staff members also attend the sessions to provide information about other CIPRA programs. Individualized follow-up is accomplished by monitoring each volunteers progress, either by direct contact with a member of the project staff during transportation to and from the tutoring site or by monthly phone calls. In the fall of 1985, the project implemented a data base to computerize volunteer files. This system provides instant access to current information on all volunteers and the operation of all programs. The system is an efficient means for organizing and managing complex program operations, such as scheduling participants into 17 driving shifts a week and keeping track of individual volunteers within the various programs. CONCLUSION The D.C. Schools Project is currently in its fourth year of operation at Georgetown University. Floretta Dukes McKenzie, Superintendent of Schools for the District of Columbia (personal communication to Father Bradley, September 4, 1986), recently lauded the program as successful in helping the students improve their English, and consequently, their classroom performance, and in helping them gain self-confidence and remain in school (p. 1). On a project evaluation form, Mike Bell, an ESL teacher at Francis Junior High School, stated that the difference in the students progress is always noticeable and often dramatic. To rate participants progress in a formal, quantitative manner, Georgetown is developing a monthly evaluation system to gather data from teachers and counselors. The program is unique in bringing together a university and a public school system to attack a grave and ongoing problem for a special population of underprivileged immigrant students and for the schools in general, which are strapped by limited resources. Moreover, it is a program that can be duplicated on two important levels. First, the volunteer force can be increased in size by initiating similar programs at other area colleges and universities that also work with the D.C. Public

School System. The Rev. John P. Whalen, Executive Director of the Consortium of Universities of the Washington Metropolitan Area (personal communication, January 27, 1987), has voiced support for expanding the program within that group. Steps have already been taken to bring student leaders from area universities to observe and participate in the Georgetown programs as a prelude to establishing their own programs. Second, the program can be duplicated in other similarly impacted areas across the nation. Following a recent report (NCBE Forum, 1987) of the program, the D.C. Schools Project has received indications of interest and requests for information from school teachers, universities, and public school districts. The D.C. Schools Project and Georgetown University, also recognizing the value of the program to the educational growth process of their participating students and alumni, stand prepared to maintain and improve their commitment to assisting the D,C. Public School System and immigrant students, as well as to offer assistance to those institutions that wish to establish similar programs. REFERENCES
Guillermoprieto, A. (1983, October 2). Salvadoran refugees straining D.C. public schools. Washington Post, p. 1. NCBE Forum staff. (1987). Programs that make a difference. NCBE Forum, 10 (2), 3.

Authors Address: D.C. Schools Project, Center for Immigration Policy and Refugee Assistance, Box 2298, Georgetown University, Washington, DC 20057



The TESOL Quarterly invites commentary on current trends or practices in the TESOL profession. It also welcomes responses or rebuttals to any articles or remarks published here in The Forum or elsewhere in the Quarterly.

Comments on Elsa Roberts Aurbachs Competency-Based ESL: One Step Forward or Two Steps Back?
A Reader Reacts. . .
Pacific Grove, California

People engaged in education and people engaged in training seldom understand each other or each others approaches, methods, and techniques. No better evidence of this has recently been published than Elsa Roberts Auerbachs recent article in the TESOL Quarterly (Vol. 20, No. 3, September 1986). I doubt that anyone will dispute her motives in arguing for educational goals for adult learners of English, but Auerbachs arguments are subtly chauvinistic. Also, she does not seem to understand that competency-based instruction in general and competency-based education (CBE) in particular are not educational programs at all, but training programs. The distinction is complex and has been widely debated, but she cites one pivotal point in the distinction when she quotes Stenhouse: Education as induction into knowledge is successful to the extent that it makes the behavioral outcomes of the students unpredictable (p. 420). Exactly! The goal of training, on the other hand, is to make the behavioral outcomes predictable! Most of what goes on in universities is not education, but trainingtraining that is requisite to the later achievement of educational goals. To cite an example of which I have personal knowledge, students of speech therapy must first be trained in several skills, including the ability to use the International Phonetic Alphabet to discriminate and record misarticulations. (ESL teachers would profit from it, too.) That is pure training, but essential for the subsequent (educational) courses in communication

therapy. In training programs, we know fairly precisely what we want the learner to be able to do and can devise very specific and objective evaluation devices to determine when the necessary skills have been achieved. So, although she does not say so, the questions addressed in Auerbachs article are what adult learners of English need: training, education, or both. The answer seems obvious to me. They need both! But when they arrive, their first need is usually training. They must become economically viable as quickly as possible; that is certain, unless they are to live with poverty and loss of pride. Then, in order to partake fully of the American culture and economy, they must have education. But does every one of them need to be educated in the United States? Auerbach seems to fall into the same trap for which she severely criticizes CBE: the assumption that they must be educated here, that is, taught Western ways of thinking. Many adult ESL students have brought with them all the education they need. What they lack is training in the basic English skills of survival until they can put their education to work for themselves and the nations economy. But these criticisms would not have been sufficient to cause me to write this rebuttal. The truly fundamental flaw in Auerbachs reasoning is that she grandly generalizes against the CBE approach because she has found flaws in mostperhaps allthe existing CBE programs. Several years ago, when I was active in EFL/ESL, I looked into some of the CBE programs and found none that satisfied me, either, none that consistently manifested the approach that CBE espouses. So you might think that I agree with Auerbach. I do not! The proposed CBE approach is still valid. It is only that the CBE programs need more persons competent to design and implement thempeople, perhaps, who have Auerbachs skills and perception and who understand the separate and vastly different roles of training and education. Furthermore, Auerbach seems to assume that educational programs, including educationally oriented ESL programs, actually teach students independent thinking. Some do, but it is the grossest of generalizations to imply that they all and necessarily do. Many educational institutions do not achieve this goal even with their students who are native English speakers, let alone students with limited English. It is certainly not uncommon for educators to look down on training as a lesser activity. I have engaged in both and find them equally challenging and equally rewarding, though in quite different ways.

One of the most satisfying aspects of working in an EFL program which is clearly identified as a training program is that it operates to a proficiency criterion. Both instructor and student know fairly exactly when the student has reached the degree of proficiency necessary to progress to the next stage of trainingor of education. If a trainer has learned from experience that a particular type of student will require 16 weeks or 20 weeks (or whatever) of intensive, 6-hours-a-day English training, then that is what he or she will get. In my experience, at least, the so-called educational programs give instruction for fixed periods of time and hope for the best. But when the purpose of education in ESL is made ambiguous to the extent described in Stenhouses perceptive statement (above), the educator can never be quite sure when that best has been achieved. Auerbachs article seems to advocate this ambiguity and even to criticize a navigator for deciding precisely on his or her destination. I sense, however, that she has the capability of rethinking the matter.

The Author Responds.


ELSA ROBERTS AUERBACH University of Massachusetts at Boston Francis Cartier makes a valid point when he argues that competency-based adult education (CBAE), despite its name, is not education at all, but rather training. This is precisely what I am critiquing about CBAE as an approach to adult ESL. There are several problems with the focus on training rather than education for adult ESL students. A training orientation, focusing on predictable outcomes, reduces language learning to a mechanistic, behavioral task which separates language from thought and from the creation of meaning. While training may be appropriate for the transfer of skills (e.g., in speech therapy), equating language acquisition with the acquisition of a sequence of skills is reductionist at best, running counter to recent second language acquisition theory and to adult learning theory. Nevertheless, Cartier implies that this approach is valid for early stages of language acquisition and for certain types of students. Are we, then, to reserve creative, communicative language acquisition

environments for higher levels of acquisition and for select students? Second, the training orientation equates language proficiency with employability. This narrow definition, which sees proficiency as merely the language necessary for limited types of interaction, may guarantee that students never gain access to education. Cumminss (1981) distinction between basic interpersonal communication skills and cognitive academic language skills may be analogous here. He argues that in bilingual education, all too often proficiency is equated with ability to interact in face-to-face, context-embedded situations and that bilingual education stops before students have a chance to develop context-reduced language proficiency. Particularly in the current funding climate, there is a similar danger in viewing adult ESL as training for employment: All too often students are passed from programs when they are seen as employable and they never get to education. Third, there is no evidence to support the view that training in fact leads to employment. As Tollefson (1986) points out, there is no correlation between successful performance on language tests and successful resettlement (as measured by rates of employment versus dependence on public assistance). Clearly, factors other than ability to communicate appropriately in specified situations mediate who gets what kind of job. What Cartier misses is the key argument of my articlethat the kind of language teaching we provide reflects a stance with regard to the broader socioeconomic context. As Skutnabb-Kangas (1979) has shown, language programs or policies which are justified by the argument that they promote rapid economic assimilation may in fact serve to reproduce existing class relationships. In the case of Sweden, she showed how just such a policy perpetuated the existence of an economically necessary underclass. The argument of my article is that we need to question whether, under the guise of training for economic access, we are not in fact serving very similar needs. Muller and Espenshade, in The Fourth Wave: Californias Newest Immigrants (1985), make a strong case for the argument that the new immigrants are vitally necessary to the economy precisely because they do jobs for which there is no other source of labor. This leaves us as ESL professionals with some hard questions: What is our role? Is it enough to provide training for immediate access to entry-level jobs? Bluntly put, are we being pressured to become agents in producing a new source of cheap labor? How can the transition between training and education take place?

This leads to Cartiers final point about the implementation of CBAE. He argues that we should not throw out the baby with the bath water: Just because so few people can put CBAE into practice correctly does not mean the approach is flawed. I would shift the blame from practitioners of CBAE to the context it takes place in. Perhaps it is not that those trying to implement CBAE are incompetent, but rather that the approach itself has inherent contradictions that make it difficult to implement. Teachers are caught in a bind of wanting to strive for broadly defined language proficiency but having to focus on language/skills for employment, given the current climate of funding. The focus on prespecified proficiency criteria may not fit the complexities of real-world demands on students. Questions about CBAE will not disappear by changing terminology or by blaming those who implement it. We need to look at the debate about CBAE in the same way that primary and secondary educators see the debate about bilingual educationas one of language policy.

Cummins, J. (1981). The role of primary language development in promoting educational success for language minority students. In Schooling and language minority students: A theoretical framework (pp. 3-50). Los Angeles: California State University, Evaluation, Dissemination and Assessment Center. Muller, T., & Espenshade, T.J. (1985). The fourth wave: Californias newest immigrants. Washington, DC: The Urban Institute Press. Skutnabb-Kangas, T. (1979). Language in the process of cultural assimilation and structural incorporation of linguistic minorities. Rosslyn, VA: National Clearinghouse for Bilingual Education. Tollefson, J.W. (1986). Functional competencies in the U.S. refugee program: Theoretical and practical problems. TESOL Quarterly, 20, 649-664.



and review articles, which discuss materials in greater depth than in a typical review, are particularly welcome. Reviews should generally be no longer than 5 double-spaced pages, although comparative reviews or review articles may be somewhat longer. Submit two copies of reviews to the Review Editor: Polly Ulichny English Department University of Massachusetts/Boston Harbor Campus Boston, MA 02125 Brief Reports and Summaries. The TESOL Quarterly also invites short descriptions of completed work or work in progress on any aspect of theory and practice in our profession. Reports of work in the areas of curriculum and materials development, methodology, teaching, testing, teacher preparation, and administration are encouraged, as are reports of research projects of a pilot nature or which focus on topics of specialized interest. In all cases, the discussion of issues should be supported by empirical evidence, collected through either formal or informal investigation. Although all reports and summaries submitted to this section will be considered, preference will be given to manuscripts of 5 double-spaced pages or fewer (including references and notes). Longer articles do not appear in this section and should be submitted to the Editor of the TESOL Quarterly for review. Send two copies of reports and summaries to the Editor, Brief Reports and Summaries: D. Scott Enright Department of Early Childhood Education University Plaza Georgia State University Atlanta, GA 30303 The Forum. The TESOL Quarterly welcomes comments and reactions from readers regarding specific aspects or practices of our profession. Responses to published articles and reviews are also welcome. Contributions to The Forum should generally be no longer than 5 double-spaced pages. Submit two copies to the Editor of the TESOL Quarterly at the above address. 2. All submissions to the Quarterly should conform to the requirements of the Publication Manual of the American Psychological Association (Third Edition), which can be obtained from the Order Department, American Psychological Association, 1200 Seventeenth Street, N.W., Washington, D.C. 20036. The Publication Manual is also available in many libraries and bookstores. 3. All submissions to the TESOL Quarterly should be accompanied by a cover letter which includes a full mailing address and both a daytime and an evening telephone number. 4. Authors of full-length articles should include two copies of a very brief biographical statement (in sentence form, maximum 50 words), plus any

special notations or acknowledgments which they would like to have included. Double spacing should be used throughout. 5. The TESOL Quarterly provides 25 free reprints of published full-length articles and 10 reprints of published reviews. 6. Manuscripts submitted to the TESOL Quarterly cannot be returned to authors. Authors should be sure to keep a copy for themselves. 7. It is understood that manuscripts submitted to the TESOL Quarterly have not been previously published and are not under consideration for publication elsewhere. 8. It is the responsibility of the author(s) of a manuscript submitted to the TESOL Quarterly to indicate to the Editor the existence of any work already published (or under consideration for publication elsewhere) by the author(s) which is similar in content to that of the manuscript. 9. The Editor of the TESOL Quarterly reserves the right to make editorial changes in any manuscript accepted for publication to enhance clarity or style. The author will be consulted only if the editing has been substantial.